Wheat in Crisis: Variability of Wheat Area in Lebanon 2017-2025

Wheat in Crisis: Variability of Wheat Area in Lebanon 2017-2025 | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Wheat in Crisis: Variability of Wheat Area in Lebanon 2017-2025 Mariam Ibrahim, Ghaleb Faour, Michel Le Page, Marielle Montginoul, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8649264/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Wheat stands as one of the most important staple crops worldwide. However, the vital role of this crop has been increasingly challenged in Lebanon, in recent years by multi-factorial crises from socio-economic, political, security and climate factors, threatening agricultural stability and food supply. Consequently, monitoring wheat production is crucial for managing import and export activities, developing effective policies, achieving resilient agricultural development, and ensuring food security. This study provides the first national, multi-year monitoring of wheat area in Lebanon (2017–2025), linking satellite observations with crisis impacts to support food security planning. We conducted a multi-temporal supervised classification from 2017–2018 to 2024–2025 seasons, using Google Earth Engine, employing Sentinel-2 optical images and Random Forest classifier. We estimated wheat area based on a random stratified sample achieving an overall accuracy of 87%. Interannual changes were then related to major crises and input-price dynamics. Wheat area increased during 2019–2021 but dropped sharply in 2021–2022 as subsidies weakened and input costs surged. Indeed, during the transition toward economy dollarization, the computed indicator of production cost expressed in USD peak in 2021–2022 and then ease consistently with the 2022–2023 area rebound reaching the highest level observed during the study period. In 2023–2025, the crop area decreased again dramatically (-34% in 2023–2024 and − 38% in 2024–2025) in relation to the conflict with Israel and associated widespread displacement of population that likely constrained field access and reduced sowing, particularly in southern Lebanon. Our findings also point out the need of ground-truth data for accurate area estimation, as well as additional information on yield and socio-economic conditions to better understand the interannual variability of wheat area. Earth and environmental sciences/Climate sciences Earth and environmental sciences/Environmental sciences Earth and environmental sciences/Environmental social sciences Wheat Lebanon Crisis Crop Mapping Sentinel-2 Random Forest Food Security Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 1. Introduction Wheat is the most critical staple crop in the global food system, and one of the most politically significant commodities in the world. Wheat domestication was founded in the Fertile Crescent of the Near East, near the Tigris and Euphrates rivers, specifically in what is now southeastern Turkey and northern Syria, where its wild progenitor was found (Lidon et al., 2014). Today, wheat is cultivated across the globe, covering over 220 million hectares worldwide. It is also the top produced cereal in Lebanon according to the Food and Agriculture Organization (FAO) statistics (FAO, 2023a). Local production covers about 20% of Lebanon’s total wheat demand, while the 80% imports are provided by Ukraine (80%) and Russia (16%) (FAO, 2023a; World Bank Group, 2022) (Fig. 1 ). Relying on imports means a high sensitivity to any external disruptions, leading to a lack of resilience and stability. Therefore, in times of crisis, the situation can be concerning. Indeed, Lebanon represents a particularly acute case study of agricultural vulnerability under compounding crises. The economic breakdown began in October 2019 (Makdisi & Amine, 2022), and was marked by currency devaluation, hyperinflation, banking sector failure, and the deterioration of essential infrastructure and services. Lebanon was hit hard by the COVID-19 pandemic soon after (Koweyes et al., 2021). Just as the country was struggling to recover, the devastating Beirut port explosion in 2020 destroyed key grain silos (UN Office for the Coordination of Humanitarian Affairs, 2020). Then, the conflict between Russia and Ukraine, the two major wheat import sources, disrupted global wheat markets (Abay et al., 2023; Devadoss & Ridley, 2024). Additionally, since October 2023, the Israeli attacks on Lebanon hit main wheat production areas (CNRS-L & UNDP Lebanon, 2024). All of this unfolded under the looming impact of climate change (IPCC, 2023). These intersecting crises produced a multifaceted challenge for the Lebanese agricultural sector, and in particular wheat production. Food security in Lebanon has deteriorated dramatically, with the country facing unprecedented levels of food insecurity across multiple dimensions: availability, access, utilization, and stability (FAO, 2006). Against this backdrop of crisis, monitoring wheat production becomes particularly relevant for understanding the challenges and opportunities within the Lebanese agricultural sector (FAO, 2023b). By accurately tracking the spatial and temporal dynamics of wheat cultivation, such a system provides the foundation to link production patterns with input costs, macroeconomic crisis indicators, and strategic food security considerations. Linking wheat production outcomes to these factors, enables valuable insights into how well the local production systems are performing and where adjustments may be needed and allows policymakers and stakeholders to design and implement more targeted and effective interventions, to enhance domestic food production capacity, optimize resource allocation and strengthen farmers’ resilience under challenging conditions. Unfortunately, the Lebanese agricultural sector faces significant challenges, complicating wheat production monitoring. Opportunistic farming and fragmented land ownership, with 70% of farmers cultivating less than one hectare, lead to unpredictable production patterns and diverse farming practices and field conditions (Caiserman et al., 2019; Dal et al., 2021). The lack of continuous and up-to-date data on wheat areas, with the last comprehensive agricultural census conducted in 2010, further undermines any baseline for effective production monitoring (Ministry of Agriculture & FAO, 2010; Wang et al., 2015). While these challenges are significant, they do not preclude the crucial need for monitoring system. The traditional crop fields identification survey, while offering high accuracy, present significant limitations including high costs, time-consuming fieldwork, and limited accessibility to certain areas. Thus, in practice, mapping becomes a compromise between the ideal and the feasible. Remote sensing emerged as transformative solution for these constraints, offering spatially and temporally cost-effective data for land-cover and land-use information derivation. Whether performing a simple plant count (Oh et al., 2020) or going further to estimate yield (Khabba et al., 2020), detect diseases (Bhattarai et al., 2019), manage irrigation (Bwambale et al., 2022), or monitor agricultural areas during wartime (Wagner et al., 2026), remote sensing technologies have evolved, particularly with the availability of Sentinel-2 (S2) imagery with 10-20m spatial resolution (European Space Agency, 2021), a substantial improvement over earlier coarse resolution (300m − 1Km) systems which often result in mixed pixels that misrepresent land cover, especially for small and locally managed land parcels (Giri et al., 2013). The integration of readily accessible cloud-computing platforms like Google Earth Engine (GEE) which offers extensive data catalog and tools for application development (Google for Developers, 2025), combined with the advancements in machine learning and deep learning algorithms, has potentially improved the process. Random Forest (RF) classifier has demonstrated strong performance in classification and is considered more user friendly, compared to commonly used classifiers, including Maximum Likelihood (ML), Support Vector Machine (SVM), Artificial Neural Network (ANN) (Liu et al., 2018; Zhao et al., 2024), and temporal convolutional neural network (Antonenko et al., 2024). Research across the Middle East and North Africa (MENA) region, demonstrates the effectiveness of these approaches. In Tunisia, combining Sentinel-2 images with 15 vegetation indices and RF algorithm achieved an overall accuracy ranging from 85.8% to 95.1% (Khlif et al., 2023). Morocco’s CerealNet, a hybrid deep learning architecture, combining a Long Short-Term Memory (LSTM) branch for phenological features extraction from the Normalized Difference Vegetation Index (NDVI) time series with a Convolutional Neural Network (CNN) branch for spectral features analysis, achieved a categorical accuracy of 95% (Alami Machichi et al., 2022). In Turkey, Sentinel-1 data, using coherence data alone yielded the highest accuracy of around 81% in crop classification (Narin et al., 2021). In Lebanon, two NDVI-based methodologies were applied specifically to the Bekaa plain: the Simple and Effective Wheat Mapping Approach (SEWMA), which exploits wheat's unique phenological patterns to distinguish it from barley and triticale, achieved 82.6–87% accuracy (Nasrallah et al., 2018) while Euclidean distance classification of NDVI profiles from 386 ground-referenced fields reached 84.06% accuracy for wheat identification (Caiserman et al., 2019). While these findings demonstrate the effectiveness of remote sensing in agricultural monitoring, and particularly, in Lebanon, yet, these applications were conducted prior to the multi-dimensional crisis and did not account for the combined effects of economic, political, and socio-environmental shocks on wheat production, leaving a critical gap in understanding production dynamics under extreme conditions. Building on this gap, our main objective is to study the variability of wheat area in the context of the multiple crises that occurred in Lebanon. In the subsequent sections, the study area and data together with the methodological materials for the classification and the assessment of uncertainties are described ( 2 ). Followed by the results section ( 3 ) which is organized into three subsections on the classification accuracy of the maps and the associated uncertainties, as well as the interannual variability of wheat area. The discussion section ( 4 ) is also organized into two subsections where we discuss the wheat mapping approach and the link with the multi-crises in Lebanon. The conclusion section ( 5 ) summarizes the main findings of the paper and opens into future investigations. 2. Material and Methods 2.1. Geographical Description and Agricultural Background of the Study Area Lebanon is situated on the eastern shore of the Mediterranean Sea, positioned between latitudes 33° and 35° N and longitudes 35° and 37° E (Fig. 2 ). Covering an area of 10,452 Km², Lebanon is predominantly mountainous with alternating geographical features. Landforms, climate, soils, and vegetation can vary significantly over short distances. The country is defined by four main geomorphological units: a narrow coastal plain, two mountain ranges (Mount Lebanon and the Anti-Lebanon) considered as the water tower of the country, and a fertile depression known as the Bekaa Valley, which lies at an altitude of 700 to 1,100 meters. In addition, there is the northern Akkar plains which are of interest for this study. Lebanon experiences a Mediterranean climate (Csa according to the Köppen-Geiger classification) characterized by four distinct seasons: a hot and dry summer, a cool and rainy winter, and moderately dry fall and spring seasons. The coastal and mountainous areas receive abundant rainfall, up to 850 and 1,100 mm respectively, mostly between December and March. In contrast, the Bekaa Valley experiences a semi-arid to continental climate with unpredictable rainfall. Precipitation decreases from south to north, and its irregularity increases along this gradient (Sanlaville, 1963). The southern Bekaa Valley, has a sub-humid Mediterranean climate for example, and receives 500 to 800 mm of precipitation annually, while the north-eastern part is almost arid to continental, as it is shielded from the sea’s influence by a high mountain ridge exceeding 3,000 m in altitude and gets less than 600 mm. At very high elevations, snow patches can persist for 6–9 months. In contrast, periods of total drought can occur, making complementary irrigation necessary for any form of intensive cultivation (Darwish et al., 2012; Fayad et al., 2017; Verdeil et al., 2016). Agriculture plays an important role in Lebanon. Spanning over 200,000 ha, the country has a relatively diversified agricultural land, including fruits, vegetables, cereals, grapes, and olives (Dal et al., 2021; Maddah & Darwish, 2023). Lebanon has three main wheat-growing regions: the north, the south, and the Bekaa Plain. Among these, the Bekaa Plain is the largest contributor (Caiserman et al., 2019), accounting for an average of 58% of the total harvested area. Indeed, the Bekaa Plain offers the ideal environment for wheat cultivation and is often called Lebanon’s breadbasket. Stretching between the slopes of the Lebanon and Anti-Lebanon mountains, the Bekaa Plain features fertile alluvial soil, providing a rich nutrient base that supports wheat growth. With cold winters and hot summers, the crop benefits from low temperatures during the vernalization phase (Acevedo et al., 2002) and higher temperatures during heading, anthesis, and grain filling (Khan et al., 2020). In general, winter wheat is sown in autumn, enters dormancy in winter, begins its green-up phase in early spring, grows rapidly until the heading phase in late spring and reaches full maturity in early summer (Caiserman, 2020; Nasrallah, 2019). It’s hard to define generalized wheat cultivation practices, as it depends on each region and each farmer means. However, in general, the season begins with plowing the land, followed by sowing the grains, and then leveling the soil (A. Abdallah, personal communication, March 6, 2025). The farmer usually fertilizes in two stages: in February, nitrogen fertilization through ammonium fertilizer is used, and in April, urea is applied. Depending on his means, the farmer may use only urea or only ammonium. Very few can afford to use potash fertilizers or 20-20-20 (20% nitrogen, 20% phosphorus, and 20% potassium) during irrigation. Some farmers avoid using synthetic fertilizers, mainly because potatoes, often grown as the preceding crop (generally cultivated from early spring to late summer of the previous year) leave behind sufficient soil fertility to meet wheat’s nitrogen requirements (Nasrallah, 2019). Despite being a winter crop, wheat often receives complementary irrigation in early spring to improve yields. In the Bekaa plain, over 80% of wheat plots are complementary irrigated, with the frequency depending directly on the rainfall received during the season. Since rainfall usually stops in February-March, farmers typically begin irrigating in mid-April and continue until mid-May. Some wheat plots, however, remain rainfed due to lack of access to water, financial constraints, or reliance on the hope of a good rainfall season (Nasrallah, 2019). Regarding weed control, 2,4-D is used, and only a few uses a product specifically targeting wild oat. Concerning rust, which may affect wheat due to humidity, azoxystrobin, penconazole, or fosetyl-aluminum may be sprayed, once rust is observed (A. Abdallah, personal communication, March 6, 2025). Farmers usually follow crop rotation seasons that include wheat, potato, corn and vegetables meaning that crop types in the agricultural areas are highly changing from one year to another. In the Bekaa valley, wheat-potato rotation is the most followed, compared to other type of rotations (Nasrallah, 2019). 2.2. Agro-Economic Context and Macro-Economic Data The timeline of our study coincides with a cascade of overlapping and sequential crises that together have placed enormous stress on the agriculture sector since 2019. These multiple crises impacted the production costs and limited the field access for the farmer. This complex situation has led to direct macroeconomic impacts, most notably rising prices, and, consequently, to a gradual adaptation by farmers that has profoundly reshaped cereal cultivation in the country. A previous study has shown that wheat was chosen primarily for its profitability, with most farmers selling their production to the government, as part of the subsidy program (Caiserman et al., 2019). The economic crisis alone, led to ( 1 ) the suspension of the subsidy program, ( 2 ) a severe currency depreciation, ( 3 ) banking restrictions reduced access to credit for farmers, ( 4 ) a rise in prices of fuel, seeding and fertilizers due to local inflation and to a lesser extent to the global market. These multiple crisis-related impacts led to a drastic increase in production costs, which in turn probably resulted in a decline in cultivated areas as anticipated by FAO (2020). For instance, nitrogenous fertilizers prices peak in 2022. Within this context, many producers consequently reduced input use or shifted to low-input crops (Khafagy et al., 2022). The 2021 removal of fuel subsidies further raised irrigation, transport and mechanical tillage costs while border hostilities in southern Lebanon since late 2023 have displaced rural populations. The covid crisis limits access to fields for farmers. The 2020 port explosion compounded pre-existing supply-chain and currency constraints affecting imported inputs such as seeds. FAO reports indicate that seeds and other imported inputs remained available but at elevated prices in late 2020. The consequences for wheat cultivation are still poorly documented. Some studies highlighted that smallholders often abandoned cereal cultivation and migrated to cities in search of income (Arafeh & Sukarieh, 2023). Medium-sized farms reduced fertilizer and pesticide applications or switched to rainfed barley, legumes, or fallow systems. Larger farms adjusted cropping patterns or invested in limited mechanization consolidation. By lack of information on farmer adaptation strategy, quantitative assessment of the impact of crisis on cereal cropping area will focus on the macro-economic consequences for the farmers and, in particular, to the production costs. Fuel prices was collected from IPT Group (IPT Group, 2025), while average fertilizer prices were sourced from the Lebanese Customs website (Lebanese Customs, 2024). Seeds prices were provided by Maison de d’Agriculteur (A. Abdallah, personal communication, March 6, 2025). Fertilizer and seed prices are provided in US dollar per ton while fuel is in Lebanese pound per 20 liters. All data correspond to an average at the country scale and are provided on a yearly basis due to the lack of a complete infra-annual price series (Fig. 3 ). From these datasets, a simplified indicator of crop production cost per hectare of wheat accounting for fertilizers, seeds and fuel for mechanical tillage operations was computed based on representative values for wheat crop (per hectare) of 120 Kg of seeds (Bashour et al., 2016 for Mediterranean wheat systems), 80 Kg of nitrogen (Boaretto et al., 2000; FAO, 1971) and ~ 65 L diesel.ha⁻¹ (order-of-magnitude operational fuel use for wheat production following Safa et al., 2010). To best reflect the local conditions faced by farmers in Lebanon, all prices were also converted into Lebanese pounds using an annual average parallel-market exchange rate compiled from multiple online sources: 1,507 LBP/USD for the pre-crisis period (until 2019), 8,900 LBP/USD in 2020, 20,000 in 2021, 35,000 in 2022, 110,000 in 2023 and 90,000 in 2024 and 2025 (in the absence of values for 2025, we assumed a stable average exchange rate reflecting the post-crisis dollarized environment). Finally, we also expressed the production costs in bread-equivalent units to make the numbers easier to interpret by relating them to the price of a basic staple. To this objective, the average annual prices of a bread bundle were extracted from the Lebanon Market Monitor report produced by the World Food Program (WFP) accessible at https://dataviz.vam.wfp.org/the-middle-east-and-northern-africa/lebanon/reports?current_page=1&country=lbn . 2.3. Ground Truth and Reference Points As the focus is put on wheat crop areas, the classification scheme was intentionally limited to five major land-cover classes: wheat, bare soil, built-up, forest (including natural forest and agroforestry fields), and vegetation gathering all type of annuals, excluding wheat). This simplification reflects both the heterogeneous nature of Lebanese agricultural landscapes and the study’s single-crop focus. Previous research has shown that limiting the number of non-target vegetation classes enhances separability and classification accuracy in phenology-based crop mapping (Immitzer et al., 2016; Liu et al., 2018; Nasrallah et al., 2018). Moreover, in a crisis context with restricted field data availability, a reduced number of classes provides a balanced training dataset and greater Random Forest stability (Probst & Boulesteix, 2018). These five categories represent the dominant land types across Lebanon during the wheat season and account for nearly all spectral variability relevant to this analysis. Water surfaces and snow were excluded from the classification process (see section 2.4 below ). Reference datasets consisted of two groups: wheat reference points gathered from ground surveys, and user defined reference points for the remaining classes, collected using high-resolution images from Google Earth. With regards to wheat ground truth, the Ministry of Economy and the Ministry of Agriculture have developed economic policies to support wheat farming and to encourage its continued cultivation until the season 2019–2020. The government purchased wheat from farmers, with a payment ceiling typically based on the amount of wheat produced per unit of cultivated land (Ministry of Finance, 2012). The National Council for Scientific Research of Lebanon (CNRS-L) was officially responsible for this process, with the primary goal of accurately identifying areas that applied for subsidies. Wheat reference points were acquired from the CNRS-L database. For the crop seasons from 2017–2018 up to 2019–2020, CNRS-L digitized wheat property requests. Teams of engineers conducted field surveys across key regions to cover all plots (Faour & Abdallah, 2020). The remaining additional reference data was acquired in the 2022–2023 season. For the latter, wheat reference points were provided by the Directorate General of Cereals and Beetroot (DGCB) - Ministry of Economy and Trade, based on a field survey they conducted that year. Table 1 shows the total number of fields surveyed (including wheat and other crops). Wheat fields were extracted from CNRS-L and DGCB databases and cross-checked using Google Earth. Then, reference points were systematically selected within the mapped polygons to ensure precise delineation and assessment of wheat coverage. Table 1 Number of surveyed fields by CNRS-L and DGCB teams. Wheat Season Number of Surveyed Fields 2017–2018 3952 2018–2019 3740 2019–2020 3038 2022–2023 1427 The reference points for other land cover classes were selected manually using high resolution images from Google Earth Pro. This method is considered reliable for validation data collection (Kennedy et al., 2007). Images covering the period from April to July of each of the four seasons (2017–2018, 2018–2019, 2019–2020, and 2022–2023) were used. Built-up, bare soil and forest points are easily detected by visual interpretation. To identify non-wheat points, and to prevent collecting points of non-subsidized wheat, we focused on the end of the growing season, when wheat will either be harvested or mature (yellow), meaning that high resolution google earth images were selected after May, 15. Areas with greenness at this stage can be considered as non-wheat (provided in the supplementary material Figure S1 ). Considering that the core objective of this study is to address the crisis scenario, one of the major implications of the crisis is the reduced accessibility to fields and logistical constraints, which directly affect the ability to collect reference data. For this reason, we intentionally limited the reference datasets to 550 points per season (110 points per class), as we also aimed to account for a potential data shortage that could emerge in the development of the monitoring system. It is important to note that, although a limited number of the available wheat fields were used to simulate a crisis context, all available wheat fields were ultimately employed to assess the classifier’s output (see section 2.5.2 ). The four datasets (Fig. 4 ) were used to train the classifier, and evaluate its accuracy. These datasets were also used as random stratified sample to estimate the wheat area accuracy, as described in section 2.5.3 (Olofsson et al., 2014). All the reference datasets were then uploaded to the GEE platform through a shape asset. 2.4. Satellite Data 2.4.1. Sentinel-2 Data Within this study, the Sentinel-2 Level-2A (S2L2A) Collection 1 was employed. As part of the Copernicus program, the European Space Agency (ESA) has developed and launched the Sentinel-2 optical imaging mission, a constellation of polar orbiting satellites Sentinel-2A, B, and C, launched in June 2015, March 2017 and September 2024. Sentinel-2 provides near real-time open access data across multiple spectral bands at 10 m, 20 m and 60 m spatial resolutions. The revisit frequency of each single Sentinel-2 satellite is 10 days while the nominal two satellites configuration provides a revisit time of 5 days. The S2L2A collection consists in orthorectified Surface Reflectance (Bottom Of Atmosphere: BOA) images with Sen2Cor processor (Main-Knorn et al., 2017). Sen2Cor performs the atmospheric-, terrain- and cirrus correction of TOA (Top Of Atmosphere) L1C input data to generate BOA, optionally terrain- and cirrus corrected reflectance images; and additionally, Aerosol Optical Thickness-, Water Vapor-, Scene Classification Map and Quality Indicators for cloud and snow probabilities (Science Toolbox Exploitation Platform, 2024). The tiling system used by ESA for S2 is a Military Grid Reference System (MGRS), based on the Universal Transverse Mercator (UTM) projection. The MGRS system divides the Earth’s surface into 60 longitudinal zones. Each UTM zone is further divided into latitude bands of 8 degrees starting from the equator. Each 6° x 8° grid cell is subdivided into tiles of approximately 109.8 Km × 109.8 Km. Adjacent tiles within the same UTM zone, overlap by around 4900 m while tiles along UTM zone borders exhibit even greater overlap with tiles from the neighboring zone (NASA, 2023). In this study, all images available covering Lebanon were considered from season 2017–2018 (September 1st, 2017 to August 31st, 2018) to season 2024–2025. Season 2016–2017 was discarded from the analysis as only Sentinel-2a images were available during a large part of the season. Details about the number of images used in each season, as well as the tiles covering the study area, are shown in Table 2 . Google Earth Engine (GEE) was used to access Sentinel-2 (S2) corrected images. Table 2 Number of images used in each season and the corresponding tiles covering the study area. S2L2A Collection Size Season Total Images Images with Cloud Percentage ˂ 10% 2017–2018 542 263 2018–2019 659 277 2019–2020 662 342 2020–2021 663 427 2021–2022 676 371 2022–2023 650 298 2023–2024 658 316 2024–2025 835 423 S2 Tiles 36SXB-36SXC-36SYB-36SYC-36SYD-37SBS-37SBT-37SBU 2.4.2. Sentinel-2 Image Pre-processing A general workflow of this section is presented in Figure S2 (supplementary material). The Sentinel-2 images were stacked to be centered on the wheat crop season from first September to end August of each season. The region of interest (Lebanon) was clipped. Sentinel-2 Cloud Probability dataset was used to mask pixels with cloud probability higher than 65% and a linear temporal interpolation was applied to gap-fill the timeseries. Snow and water mask were applied, using the Normalized Difference Snow Index (NDSI) and the Normalized Difference Water Index (NDWI) respectively as follows: \(\:NDSI\:>0.4\:\:\:\:\:\:;\:\:\:\:\:\:\:\:\:NDSI=\frac{RGB-SWIR1}{RGB+SWIR1}\) (Hall et al., 1995) \(\:NDWI\:>0.5\:\:\:\:;\:\:\:\:\:\:\:\:NDWI=\:\frac{Green-NIR}{Green+NIR}\) (McFeeters, 1996) Blue, Green, Red, Red-Edge1, Near Infra-Red (NIR) and Short Wave Infra-Red (SWIR)1 and SWIR2 bands were selected along with the Normalized Difference Vegetation Index (NDVI) (Rouse et al., 1973) to create the input for the classification’s method. NDVI is a chlorophyll sensitive index, widely employed to quantify plant biomass and net primary productivity, to highlight vegetated content from other land cover types and to compare seasonal changes in vegetation growth (Myneni & Williams, 1994). Green, and Red-Edge bands correlate with chlorophyll and other leaf pigments (Delegido et al., 2011; Fernández-Manso et al., 2016), whereas SWIR is sensitive to water content and vegetation structure (Braga et al., 2021; Ceccato et al., 2002; Hunt & Rock, 1989). Previous studies also emphasized the importance of SWIR and Red-edge bands in classifying vegetation (Chaves et al., 2020). Sothe et al., (2017) identified these bands as decisive attributes to differentiate similar phenology. Furthermore, these bands were considered most informative for vegetation classification (Macintyre et al., 2020). Finally, the median values were computed on a monthly time scale to observe the temporal dynamics of the classes while avoiding GEE memory-limit issues. For each season, the dataset consists of 12 median monthly images, each with 8 spectral features (NDVI plus the 7 selected bands), resulting in a total of 96 features. 2.5. Image Classification and Assessment 2.5.1. Classification Algorithm The general workflow of the image classification is presented in Fig. 5 . Concerning the classification algorithm, a Random Forest (RF) was used. RF aggregates multiple decision trees (Breiman, 2001) and each decision tree casts a vote for a class label, and the forest outputs the class with the highest number of votes. This aggregation mechanism is referred to as majority voting (Probst & Boulesteix, 2018). Two parameters need to be defined when using RF: the number of features used at each node to generate a tree, denoted as ‘m’ and the number of decision trees, denoted as ‘T’. In our study, ‘m’ was set to the square root of the total number of features. Reducing ‘m’ weakens individual trees but lowers their correlation, which in turn, strengthens the forest overall accuracy. RF uses bagging or bootstrap aggregating, meaning it creates training datasets by randomly resampling the original data with replacement, so that each tree sees a different version of the data. However, this means that some features may appear multiple times, while others might be underrepresented and not appear at all (Rodriguez-Galiano et al., 2012). A study showed that the biggest performance gain in RF happened within the first 100 trees, however, it is important to note that other parameters can also influence the results, such as the sample size, the number of variables (Probst & Boulesteix, 2018). Breiman, (1996) showed that increasing ‘T’ leads to the convergence of the generalization error without overfitting, thanks to the strong law of large numbers (Feller, 1991). To ensure good coverage of all observations without excessive computational cost, a forest of 300 decision trees was constructed to classify the input image. This number was determined through a sensitivity analysis, in which 2019–2020 data were used to train a classifier with an increasing number of trees. The trained classifier was then applied to 2018–2019 season, and its performance was evaluated. The results are shown in Table 3 . Although the results are quite stable independently of the number of trees, the highest overall accuracy and kappa coefficient values were recorded for T = 300, therefore, it was selected as final choice. Table 3 Overall Accuracy and Kappa coefficient of the Random Forest classifier with increasing number of trees. Number of Trees 30 50 100 200 300 500 OA 0.89 0.89 0.89 0.90 0.90 0.90 Kappa 0.86 0.86 0.86 0.87 0.88 0.87 To assess the temporal robustness of our classification strategy, we performed crop type mapping over multiple wheat growing seasons from 2017–2018 to 2024–2025. Rather than training a new classifier for each year, we aimed to train a single RF model using one season data, which would then be applied to all other seasons. This approach is intended to mimic the production workflow for an operational monitoring service, where in situ data collection and model retraining cannot be carried out every year, particularly in situations where collecting reference points is difficult due to crises, financial or logistical constraints. By comparing the accuracy obtained with this fixed classifier to that of annually retrained models, we evaluate the expected degradation in performance over time and quantify the trade-off between operational efficiency and classification accuracy. The classifiers were trained separately on each dataset (2017–2018, 2018–2019, 2019–2020, and 2022–2023), and their accuracies were assessed on the remaining three seasons for which we had reference datasets. The 2019–2020 dataset was divided into training (70%) and testing (30%) subsets. After performing the accuracy assessment, the trained classifier was exported and used to classify the multi-temporal input images from 2017–2018 to 2024–2025. Eight classification maps for Lebanon covering the seasons from 2017–2018 until 2024–2025, at a spatial resolution of 10 m were generated. 2.5.2. Classification Assessment and Consolidation The map assessment is performed using classical metrics (confusion matrix, overall, producer and user accuracy, kappa coefficient) as detailed in the supplementary material Section S3. As an additional assessment step, the available wheat fields from the four seasons were used to evaluate how accurately the predictions made by the classifier align with the true data. The wheat parcels for each season were overlaid on the corresponding classification maps, and using the Zonal Statistics tool in QGIS, the predominant mapped class within each parcel polygon was recorded. The number of correctly mapped parcels was then calculated to determine the percentage of agreement between the classifier output and the ground truth. Several post-classification assessment methods were implemented. A general workflow of this part is presented in Fig. 6 . First, change maps and transition matrices were established to evaluate class stability, consistency, and the continuity of the maps. These matrices help understand class behavior, detect trends, and identify real-world changes. Post-Classification comparison is a method of change detection, which requires the comparison of independently produced classified images (Gordon, 1980; Howarth & Wickware, 1981; SINGH, 1989). To this objective, the pixels that remained in the same land cover class, as well as those that transitioned to a different class in the following season were identified. Transition probabilities were calculated for each class based on the class total number of pixels in each season. By identifying unlikely transitions, we assessed the classification errors behind these unreasonable changes. The two main “unlikely” transition identified is the change from forest or built-up to another class. Within this context, we used repetition across years to reinforce confident forest and built-up classifications, correcting the transitions that seemed unreasonable within these two classes. For each pixel across the 8 yearly maps, we checked how many times it was classified as built-up or forest. If a pixel was classified as built-up 4 or more times, we corrected it to built-up in all years where it had a different class. Similarly, if a pixel was classified as forest 4 or more times, it was corrected to forest in the years where it had another class. Indeed, even if new buildings are constructed or new tree plots can be planted, this will affect only a relatively small area. If a pixel was classified as built-up in one year but appeared less than 3 times across the 8 maps, it was labeled as uncertain built-up. If a pixel was classified as forest in one year but appeared less than 3 times, it was labeled as uncertain forest. The RF classifier provides the estimated probabilities for each land cover class for every pixel in each season. Within this study, we also computed and used the confidence, defined as the probability of the predicted class, and the score margin, defined as the difference between the top two class probabilities, in order to select only confident predictions, while retaining flexibility by controlling how strongly the wheat class needs to dominate over the second-highest class (based on score margin). Pixels with a confidence higher than 50% and a score margin higher than 0.3 were extracted and wheat area was calculated accordingly. It is important to note that the classifier’s confidence score represents its internal certainty in each prediction based on patterns learned from the training data. However, it does not indicate real-word accuracy, which can only be determined through independent validation. 2.5.3. Sample-based Area Estimation and Accuracy Assessment Automated classifications, especially in case of large areas, will inevitably contain some errors and biases. Errors can arise from different factors such as noise, unclear class boundaries, gaps in the time series. Biases can result from the methods used for classification and change detection, as well as from human choices when collecting training data, which are often based on available reference data. To derive unbiased area estimates and quantify classification uncertainty, we adopted the Random Stratified Sample (RSS) approach proposed by Olofsson et al., (2013, 2014). This method replaces the pixel-counting procedure, which assumes the map is error-free, by a statistically rigorous estimation of class proportions derived from a reference sample. Given that reference data does not differentiate between confident and non-confident wheat classification. The Random Stratified Sample (RSS) method was applied on the raw classification map (before selecting the confident wheat pixels) in order to estimate the errors on the estimated area of wheat crops. Then, the confident wheat area and the uncertainties were extracted from the total wheat estimated area. Overall, this method ensures that reported land-cover areas and their uncertainties are statistically unbiased, and comparable across studies, representing current best practice for area estimation in land-cover mapping. The computing details are provided in the supplementary material Section S4. 3. Results 3.1. Classification Maps Assessment 3.1.1. Statistical Metrics Table 4 reports the classification metrics for the four seasons with respect to in-situ measurements, and for the cross-season runs where a classifier trained on one season is applied to the other seasons. The classifiers show on average high performance with Overall Accuracy (OA) and Cohen’s Kappa Coefficient (Kappa) values mostly above 0.80. The model trained on 2019–2020 exhibits the highest temporal transferability, with overall accuracy values ranging between 0.90 and 0.94 and Kappa coefficients between 0.88 and 0.92 when applied on adjacent seasons 2017–2018 and 2019–2020. The good correspondence between OA and Kappa also suggests a strong agreement between predictions and reference labels beyond what would be expected by random identification. In fact, Kappa, which corrects OA for random agreement, remains close to OA values (Congalton & Green, 2019; Foody, 2002; Landis & Koch, 1977). By contrast, the classifier trained on 2022–2023 performs less consistently when transferred backward in time (OA ≤ 0.80; Kappa ≤ 0.75). Overall, the observed decrease in both OA and Kappa highlights the change of land-cover spectral characteristics and probably of land management practices as well potentially linked to an anomalously dry climate in the Eastern Mediterranean region at this period (cf. the precipitation amount for season 2022–2023 in Fig. 9 below). Table 4 Accuracy assessment metrics for the four classifiers tested. Train/Test 2017–2018 2018–2019 2019–2020 2022–2023 Classifier 2017–2018 OA = 0.96 Kappa = 0.95 OA = 0.86 Kappa = 0.82 OA = 0.87 Kappa = 0.84 OA = 0.79 Kappa = 0.74 Classifier 2018–2019 OA = 0.91 Kappa = 0.89 OA = 0.95 Kappa = 0.94 OA = 0.90 Kappa = 0.88 OA = 0.79 Kappa = 0.74 Classifier 2019–2020 OA = 0.92 Kappa = 0.90 OA = 0.90 Kappa = 0.88 OA = 0.94 Kappa = 0.92 OA = 0.80 Kappa = 0.75 Classifier 2022–2023 OA = 0.87 Kappa = 0.84 OA = 0.79 Kappa = 0.74 OA = 0.76 Kappa = 0.70 OA = 0.86 Kappa = 0.83 Table 5 displays the Producer Accuracy (PA) and User Accuracy (UA) for each class using the 2019–2020 that exhibited the best overall performance based on global statistical metrics (OA and Kappa). Overall, class accuracies show strong performance for most land-cover types across years. Wheat and built-up areas consistently exhibit high PA and UA (> 0.90 in most years). Forest also exhibits high UA (> 0.90), although PA fluctuates slightly. Bare soil shows more variability across years, likely reflecting differences in soil moisture, tillage state and vegetation residues influencing its spectral separability. The most pronounced decline occurs for the vegetation class in 2022–2023 as PA drops sharply to 0.41 and UA to 0.70, suggesting a substantial change in the spectral response or separability of this class as already observed on the global metrics above. This was also partly expected as the vegetation class gather all type of vegetation that is not classified as forest or wheat meaning the intra-class spectral variability is high by definition. Table 5 Producer Accuracy (PA) and User Accuracy (UA) for each cover class in the four seasons with reference data using the 2019–2020 classifier. Wheat Season Metric Wheat Vegetation Forest Bare soil Built-up 2017–2018 PA 0.95 0.85 0.90 0.94 0.95 UA 0.93 0.86 0.94 0.95 0.91 2018–2019 PA 0.99 0.95 0.82 0.83 0.92 UA 0.93 0.82 0.95 0.96 0.87 2019–2020 PA 0.97 0.97 0.82 0.97 0.97 UA 0.97 0.86 1.00 0.94 0.94 2022–2023 PA 0.86 0.41 0.90 0.85 0.96 UA 0.71 0.70 0.91 0.82 0.82 The analysis of compliance between the classifier and the ground truth data, using all available wheat fields (from CNRS-L and DGCB), showed that on average, 81% of the fields were correctly mapped by the classifier across the four seasons (Table 6 ). The remaining fields were generally misclassified as vegetation. This result is somewhat expected, given that UA of the vegetation class was lower than that of wheat, indicating that a larger proportion of the areas classified as vegetation did not actually belong to this class. Table 6 Compliance % between the surveyed wheat fields and the classification maps. Season 2017–2018 2018–2019 2019–2020 2022–2023 Compliance % 75 82 88 79 As a conclusion and hereafter, the classifier 2019–2020 is chosen as the best candidate for classification on the whole time series. Considering wheat that is the subject of the present study, accuracies remains very high for most the years with a slight decline in 2022–2023. Overall, these results indicate that the model trained on 2019–2020 data generalizes quite well and will be chosen hereafter as the classifiers for mapping land cover type along the whole time period. Based on this classifier, seven wheat maps for Lebanon covering the seasons from 2017–2018 until 2023–2024, at a spatial resolution of 10 m were generated. Finally, as a complementary way to assess classification accuracy, Figure S5 (supplementary material) displays the mean confidence and the mean score margin for each class across the years based on the 2019–2020 classifier. Values range between 45 and 70% depending on the class and the seasons. However, such raw values must be interpreted with caution: they are not true posterior probabilities unless the model is well calibrated (Berio Fortini et al., 2024). By contrast, the score margin (the difference between the highest vote proportion for the predicted class and the second-highest vote proportion) offers a more robust indicator of separation between classes. Empirical studies in remote sensing classification demonstrate that enforcing a minimum score margin around 0.30–0.40 markedly improves class-label reliability (Aitkenhead & Dyer, 2007). Although the lowest performance is observed for the vegetation class, our results show margins consistently exceed 0.30 for all classes proving clear separability between the top two classes 3.1.2. Transition Matrices The analysis of classes transitions over consecutive seasons reveals an acceptable degree of stability, with some changes in class distribution (Fig. 7 ). For wheat, most observed transitions involve shifts to vegetation class probably because of crop rotation. Some transitions seemed slightly unreasonable, particularly the wheat-built-up and forest-vegetation ones; therefore, further investigation was conducted to evaluate the accuracy of these probabilities. An average of 10% transition from wheat to built-up was depicted for instance, which is unusually high. This is likely due to a classification error. Similarly, forest class stood out with some anomalies, showing unexpected transitions (24% of forest transitioned to vegetation). This can highlight real forest degradation or classifier confusion. As mentioned before, the vegetation class is too broad (high internal variance). Moreover, forests in Lebanon are significantly fragmented with only 10% being over 200 ha. Natural forest associations consist of pure and mixed stands including different species, notably Cedars of Lebanon, various types of Junipers (Talhouk et al., 2001). This can create overlap between classes as observed by the confidence level and score margin from Figure S5 (supp. material) , where vegetation class exhibits the lowest values. Therefore, the 24% transition from forest to vegetation is likely inflated by this ambiguity, and could be partly artefact, not entirely pure ecological change. Within this context, the classification maps were corrected based on this unlikely transition. Table 7 reports the change before and after correcting the built-up and forest pixels in terms of unlikely transitions (%). To this objective, we analyzed the repetition of built-up and forest pixels across the seven seasons and we set a threshold of at least 4 repetitions over 8 seasons for pixels to be considered true built-up/forest pixels. After correcting the new misclassified pixels, the transition probabilities were significantly improved with drastic drops of unlikely transition: forest-vegetation going from 26% during the 2018 to 2019 transition to 3%. Likewise, the wheat-built-up was 20% from 2021 to 2022 and dropped to 5%. Table 7 Transition probabilities (%) before and after correction of forest and built-up pixels. (2018→2019 means the transition from season 2017–2018 to 2018–2019). Transition 2018 → 2019 2019 → 2020 2020 → 2021 2021 → 2022 2022 → 2023 2023 → 2024 2024 → 2025 Wheat → Built-up 9 → 1 8 → 4 11 → 2 20 → 5 5 → 2 7 → 2 19 → 5 Vegetation → Forest 8 → 4 8 → 2 6 → 3 9 → 1 5 → 2 7 → 2 2 → 1 Forest → Vegetation 26 → 3 17 → 8 21 → 9 24 → 7 32 → 6 28 → 10 35 → 8 Built-up → Wheat 8 → 2 6 → 1 5 → 2 2 → 1 11 → 3 4 → 1 2 → 0 3.2. Wheat Maps and Wheat Area Estimation For visualization purposes, Fig. 8 shows wheat pixels in a small agricultural region located in the Bekaa, between Zahle and the West Bekaa, the breadbasket of Lebanon, as an illustration of the classification maps. The complete maps are provided in Figure S6 (supplementary material). The wheat/non-wheat masks exhibit a good spatial consistency: field boundaries are well resolved and parcel geometries can be clearly observed. In addition, the interannual variability proved to be high with only a small fraction of parcels remains under wheat in two consecutive seasons. This is also consistent with local crop-rotation practices. Indeed, multiple sources explain that Bekaa farmers frequently rotate wheat with potatoes and other vegetables (Caiserman et al., 2019; Jaafar et al., 2024; Tawk et al., 2019). Wheat area fractions with 95% confidence intervals were estimated using a random stratified sample for four seasons where reference points were available. A linear trendline (R² = 0.9) was fitted to examine the relationship between the wheat fraction estimated from pixel counts and that estimated from the RSS. The trendline is shown in Figure S7 (supplementary material). The trendline equation was then used to estimate the fraction in seasons where samples were not available. For the confidence intervals, the average error rate (14%) was applied to determine error intervals for the remaining seasons. The confident wheat fraction, with 95% confidence intervals, was then derived from the estimated values based on the fraction of confident pixels in each season. Finally, the area was calculated in hectares. The comparison between the two estimation methods results (pixel counts and RSS) showed that the mapped wheat area was underestimated in all seasons except for 2018–2019 (Table 8) with a negative bias varying from − 4,272 ha (almost 14% of the wheat cropped area) to -657 ha (1%) in 2023–2024. Table 8 Wheat area based on the random stratified sample and pixel count. In grey, are reported the seasons when the linear trendline was used by lack on in situ data (see text). Season Pixel Count Area (ha) Random Stratified Sample Area (ha) Bias (ha) Bias (%) 2017–2018 25,536 29,835 -4,272 14 2018–2019 29,469 27,212 + 2,257 8 2019–2020 29,039 30,432 -1,393 5 2020–2021 53,005 53,751 -746 1 2021–2022 32,604 35,449 -2,845 8 2022–2023 66,383 69,194 -2,811 4 2023–2024 51,142 51,799 -657 1 2024–2025 35,491 37,642 -2,151 6 3.3. Interannual Variability of Wheat Area We observed that the harvested area fluctuated over the seasons (Fig. 9). Wheat area decreased by 10% from 2017–2018 to 2018–2019, then increased again at a similar rate in 2019–2020. It rose sharply by 43% in the following season (2020–2021). In 2021–2022, wheat area dropped significantly by about 52%, then increased again in 2022–2023, reaching the highest level of the study period. However, it declined once more by around 34% in 2023–2024 and continued decreasing by 38% in 2024–2025. Overall, the wheat area showed alternating rises and falls, ranging from a minimum of 27,212 ha in 2018–2019 to a maximum of 69,194 ha reached in 2022–2023, reflecting a pronounced interannual variability (coefficient of variation: 35%). The analysis of the average precipitation in the main wheat producing regions during wheat season showed no correlation (-0.01) between wheat area and precipitation (Fig. 9). Particularly, in the seasons where we observed sharp variations in wheat area (2021–2022, 2022–2023 and 2023–2024), precipitation did not vary in the same pattern or intensity. Furthermore, the 2020–2021 wheat season began with a pronounced drought, as SPEI-90d values fell sharply in September 2020 (provided in the supplementary material Figure S8 ). This severe moisture deficit persisted for several months, yet wheat cultivation area during this cycle was higher than in the previous season, despite the fact that hygrometric conditions had been more favorable in 2019–2020. Seasonal precipitation is a key driver of wheat growth, particularly early in the season when adequate rainfall replenishes soil water reserves and supports crop establishment; however, its influence in Lebanon is partly moderated because many farmers buffer irregular rainfall through supplemental irrigation. Over 80% of wheat plots are supplementary irrigated (Nasrallah et al., 2019), unlike other countries, such as Morocco, where approximately 83% of wheat is grown under rainfed conditions (Lago-Olveira et al., 2024). Farmers there, typically wait to assess rainfall levels before sowing, and may abandon planting if rainfall appears insufficient, which is not the case in Lebanon. According to a past survey, 62% of West Bekaa farmers irrigate once or twice, while the remaining farmers irrigate three or four times (Tawk et al., 2019). However, it’s important to note that although precipitation itself does not seem to influence wheat area, water availability can still indirectly shape farmers’ decisions. In addition, during periods of economic or political crisis, the feasibility of supplemental irrigation becomes more constrained: high fuel prices raise the cost of pumping groundwater, the price of purchased irrigation water may increase, and labor requirements for manually irrigated fields can become prohibitive. Under these conditions, wheat water requirements may not be fully met, consequently affecting yields (Achli et al., 2025), and potentially influencing whether farmers choose to continue cultivating wheat or shift to other crops. The macro-economic factors beyond climate that may influence harvested area are discussed in section 4.2 . 4. Discussion 4.1. Wheat Mapping Approach As already observed on the wheat mask on the Bekaa Valley (Fig. 8 ), the transition matrices showed that, on average, 22% of wheat transitioned to vegetation, from one season to another, which can be partly explained by crop rotation practices in Lebanon. Unlike monoculture, diversified rotations can increase the yield. Moreover, including legumes stimulates soil microbial activities, adds nitrogen to the soil, increases soil organic carbon stocks, and enhances soil health, thereby improving fertility for the next crop. Rotations also can help improve soil structure and prevent issues like compaction (Volsi et al., 2022; Yang et al., 2024). Other than rotation, the transition to vegetation or bare soil can also result from crises factors that may influence a farmer’s choice to cultivate or not. This was further supported by the analysis of transitions between bare soil/vegetation and wheat. Table 9 reports the wheat gain and losses during the transition between two consecutive years during the study period. Two critical seasons stood out: 2021–2022 and 2023–2024, during which we observed a sharp increase in wheat loss and a significant decrease in wheat gain, deviating from the patterns observed in the previous seasons. Conversely, in 2022–2023, wheat loss declined to its lowest level, while wheat gain increased. These results indicate that new factors emerged and influenced the farmers’ crop choices. Specifically, during 2021–2022, 2023–2024 and 2024–2025, farmers faced severe financial and political constraints, respectively. In contrast, the relative stabilization of the Lebanese economy in 2022–2023 encouraged farmers to reinvest in wheat production. These variations are further discussed in section 4.2 below . Table 9 Wheat gain and loss probabilities over the study period. (2018→2019 means the transition from season 2017–2018 to 2018–2019). Seasons Transitions 2018 → 2019 2019 → 2020 2020 → 2021 2021 → 2022 2022 → 2023 2023 → 2024 2024 → 2025 Wheat → Vegetation (%) 31 21 20 21 22 27 14 Wheat → Bare soil (%) 4 13 12 20 5 9 24 Vegetation → Wheat (%) 7 11 11 8 11 7 12 Bare soil → Wheat (%) 8 4 4 1 8 4 1 Extracting all wheat pixels identified by the classifier can include low-confidence predictions, which may lead to an overestimation of the wheat area. Conversely, selecting only pixels with a confidence level above 80% risks underestimating the actual wheat area. Moreover, as confidence and score margin levels vary, the expected wheat pixels also change (Fig. 10 ). Therefore, achieving a reliable estimate requires selecting pixels based on a combination of confidence and score margin to ensure that only reasonably confident pixels are included. By this, we mean pixels that are not selected randomly and are well differentiated from other classes. For this reason, we selected pixels with a confidence greater than 50% and a score margin above 0.3. Finally, interestingly enough, the interannual dynamics remain similar regardless of the chosen confidence level or score margin threshold, meaning, that the choice of the thresholds is not critical for this study. Area estimation based on simple pixel-counting is biased because it does not take classification errors into account; as a result, the area estimates are erroneous. Misclassified pixels lead to misestimation of class areas (Olofsson et al., 2014). Indeed, when comparing wheat area derived from pixel counts to the area estimated using the RSS approach, our results showed a systematic bias. On average, an underestimation of the wheat area was observed across all seasons. This finding reinforces the importance of using RSS-based area estimation to obtain more reliable results. The presence of errors affects the width of confidence intervals for the area estimates, the larger the errors, the greater the uncertainty. Particularly, omission error in maps can add substantial uncertainty to area estimates (Olofsson et al., 2020). In the 2018–2019 season, PA which is the complement of the omission error (Gallego et al., 2010) was the highest, which means the omission error in the wheat class was the lowest. This can explain the narrow confidence interval observed for this season (9%). For the remaining three seasons, the confidence intervals were wider. Based on the OA and PA, we expected the confidence interval of 2022–2023, to be the highest and indeed, that was the case (18%). However, our results showed that the error rate in 2019–2020 (17%) was higher than in 2017–2018 season (14%), despite the higher PA in 2019–2020. This difference can be explained by the sample size. For the 2019–2020 season, the sample represented 30% of the sample used for 2017–2018, which can explain the higher confidence interval for this season. 4.2. Link between Multiple Crises and Wheat Growing Areas Figure 11 displays the wheat areas over the whole country and associated uncertainties together with the timing of the main crisis. The pre-crisis period (2017–2018 and 2018–2019) was overall stable with wheat areas remaining around 28,000–30,000 ha. The predictable economic conditions allowed farmers to plan cultivation decisions based on established cost-revenue relationships and access inputs through functional markets. The relatively low wheat areas during this period, likely reflect structural constraints in Lebanon’s wheat sector, where wheat often remains a rotation crop competing with higher-value alternatives such as potatoes or vegetables (Tohme Tawk et al., 2019). For instance, Nasrallah et al., (2019) stated that the relatively low wheat areas during season 2016–2017 (outside of our study period) was related to the choice of farmers to plant potatoes during this season. In addition, FAO GIEWS, (2018) reported unfavorable early-season conditions during winter season 2017–2018, which could have further constrained sowing decisions and reinforced low baseline areas prior to the 2019 economic crisis. Wheat area slightly increased by 11% in 2019–2020, notably after the beginning of the financial crisis in Lebanon (Fig. 11 ). During this crop season, the Lebanese Agricultural Research Institute (LARI) distributed wheat seeds free of charge upon the Ministry of Agriculture (MoA) request. Wholesalers also provided farmers with financial resources early in the season to guarantee cash flow that the farmers would repay in-kind (as a portion of the output) at the time of the harvest (FAO, 2020). Moreover, local wheat production was still supported by the government. All these factors were still encouraging farmers to keep cultivating wheat. The following boom-bust cycle between 2020 and 2022 reveals the combined influence of strategic priorities and economic constraints on farmer decision-making under extreme crisis pressure. In 2020–2021, the expansion of wheat area by 83% with regards to the three previous year average, occurred despite three concomitant crises: the deepening of the financial collapse, the lock down due to the pandemic and the Beirut port explosion. Although, the production cost increased by 40% during this season (Fig. 12 ), the area increased, violating conventional agricultural economics where area is expected to decline in response to high costs, and highlighting that food security crisis can trigger powerful counter-cyclical responses. In fact, the destruction of Lebanon’s major grain silos in Beirut port in August 2020 created immediate threat of bread shortages (Breisinger et al., 2023), prompted coordinated government and international donor emergency interventions (FAO, USAID, World Bank). The global markets were disrupted by the COVID-19 lock down and the prices were rising in Lebanese market (food inflation rose from 11% to 205%). Food security concerns overrode economic calculations, triggering return migration from urban areas to family farms, where wheat cultivation provided food self-sufficiency, supplemental cash income, and some independence from the failed financial system. Furthermore, a new local domestic pasta manufacturer has started operations, using 100% Lebanese durum wheat. This increased the opportunity for wheat farmers to sell their production, further explaining the increase in wheat area. In 2021–2022, growing wheat became a major challenge, as the production cost became fundamentally untenable, with LBP costs exploding by 327% (the steepest single-year increase in the entire study period). Since 2021, fuel prices have been increasing, rising from 315,000 Lebanese Pounds for 20L of Diesel in 2021 to 841,000 Lebanese Pounds in 2022 (IPT Group, 2025). Furthermore, fertilizers price increased drastically. According to Lebanese Customs, the price of one ton of fertilizer, which had hovered around 253 USD from 2017 to 2020, sharply increased to 385 USD in 2021, and reached 596 USD in 2022, notably after the conflict between Russia and Ukraine began in February 2022 (Lebanese Customs, 2024) (Fig. 3 ). According to data provided by Maison d’Agriculteur, the average price of local wheat seeds has increased by approximately 22% in 2021. In contrast, the average price of imported wheat seeds has remained stable in this period, at around 1,100 USD per ton, largely constrained by the limited purchasing capacity of farmers. The 252% deterioration in LBP-USD exchange rate meant that every USD-priced input became approximately 3.5 times more expensive in LBP terms, creating a fundamental mismatch between farmer revenues and input costs which were effectively denominated in USD given import dependency. At the same time, the real burden, measured in bread bundle equivalent, expanded by 129%, while the food price inflation peaked at a hyperinflationary 440% (the highest level recorded in Lebanese history). As a result, farmers needed 2.3 times more purchasing power than the previous season to simply cover production costs, representing an impossible demand. Furthermore, the crisis was directly linked to the government’s inability to sustain subsidies, confirmed by the use of foreign reserves to subsidize vital commodities like wheat (Abou Ltaif et al., 2024). Consequently, for this season, the government was no longer able to support the farmers, causing them to lose an important marketing channel, and forcing them to find alternative market routes, capable of absorbing their high production costs, to allow even minimal profits. Due to the financial collapse, farmers also lacked access to credits to purchase new inputs and were requested to pay old arrears, and to purchase inputs in cash either at face value in US dollars or in Lebanese pounds using unofficial exchange rate, translating into an increase in agricultural input costs compared to the previous seasons, based solely on exchange rate fluctuation (FAO, 2020). This situation affects particularly, small farmers who lack sufficient capital and cannot sustain operations without immediate cash flow. Changing their agricultural practices also carries the risk of lowering their productivity, as higher yields were mainly driven by the application of improved varieties, irrigation, fertilizers, and pesticides (Tadesse et al., 2017; Tita et al., 2025). All these elements made wheat cultivation a losing scenario, which explains the sharp decrease in harvested area (52%). A remarkable recovery happened in 2022–2023, despite continuously increasing production costs. Interestingly, the observed increase in LBP-denominated costs was driven mainly by exchange rate deterioration rather than by international price inflation. Indeed, a decline in USD costs (provided in the supplementary material Figure S9 ) was noted. Furthermore, the February 2022 onset of the Russia-Ukraine war, disrupted global wheat markets and caused a spike in wheat prices (Kuhla et al., 2024). High international wheat prices made domestic wheat production economically attractive and when combined with exchange rate depreciation, amplified nominal profitability in LBP terms. Food inflation also moderated to 310.8% (still extreme but decelerating), partially restoring purchasing power and the real burden (bread bundle equivalent) declined. Indeed, in 2023, after four years of severe financial crisis, a small degree of stabilization began. Lebanese economy began partially switching to dollarization, maintaining a minimum level of stability in prices (Banque du Liban, 2023). Farmers likely started to adapt to this post-crisis environment, and optimize input choices, helping to explain the expansion of wheat area in this season. Unfortunately, as the wheat area began to increase, the offensive Israeli war on Lebanon started in October 2023. This renewed instability explains the decrease in wheat area, particularly in the south, which was the most affected region, recording the highest number of attacks (provided in the supplementary material Figure S10 ), and witnessed the highest average loss percentage of 47% and 61% in 2023–2024 and 2024–2025 seasons, respectively (Fig. 13 ). By December 2023, caretaker Minister of Public Health Firas Abiad announced that around 900,000 people were displaced (OCHA, 2024). Farmers may have abandoned their land due to displacement, or they risked to be injured or dead from shelling, airstrikes, while attempting to plant or harvest, and even though the production cost was more favorable than the previous season, farmers simply could not cultivate land in active conflict zones regardless of input costs, output prices, or profitability considerations. This is supported by the fact that the transition from wheat to bare soil increased from 9% to 24% in 2024–2025 season (see Table 10 above). Overall, based on this general analysis, the choice to grow wheat appears to result from multiple influencing factors. In time of crisis, understanding this choice becomes more complex due to the variability of the factors between regions and the diverse ways farmers think and respond. Previous observations support this point of view, suggesting that individual farmers’ decisions and responses to the changing agricultural environment, such as biophysical conditions, socio-economic factors, and policy interventions, collectively shape agricultural patterns over time. Therefore, the choice to grow a certain crop is not a simple decision but a multifaceted process, especially under risk. In Vietnam, for example, risk aversion is likely linked to staying in rice monoculture, despite the strong beliefs and desires to shift to production of other crops because of rice low profitability. In contrary, knowledge of new alternatives increase the likelihood that farmers will switch from rice monocultures to other crops (Le et al., 2024). In Ghana, 41% of farmers preferred to grow cocoa instead of food crops, because of high market risk for food crops, as the cocoa sector offers a ready market and price hikes. However, others avoid cocoa farming due to perceptions of lower potential returns compared to food crops. For them, cocoa requires higher input and labor costs, coupled with the longer maturity period, making food crop farming more profitable (Hashmiu et al., 2022). In Nigeria, violent conflicts are major factors shaping agriculture. They often reduce the amount of land harvested, largely because farmers fear imminent attacks. In addition, cropping patterns are affected, with a substantial shift from long-term crops, such as perennials, to short-term crops (Amare et al., 2025). This is the case also in Ukraine where a drop of 25% of wheat productivity was observed in the areas invaded by Russia (Antonenko et al., 2024). In the case of Lebanon, previous studies have also highlighted that the choice to grow wheat isn’t constant, but rather but rather depends on the environment in which farmers operate. Socio-economic surveys conducted in 2017, showed that major farmers in the Bekaa region chose wheat because of its profitability. Since the government purchased the entire production each year, wheat was considered a secure crop compared to others, such as vegetables, which had fluctuating and unreliable prices from one season to another (Caiserman et al., 2019). Furthermore, Nasrallah et al. (2018), attributed the decrease in wheat area from 2016 to 2017 to various factors, including agricultural practices, corrupt subsidy policies, the Syrian war, and marketing policies. Several reports have investigated the state of agriculture in Lebanon and the adaptation techniques adopted by farmers, such as decreasing expenses for agricultural inputs (United Nations, 2023) by choosing substitution for cheaper alternatives (replacing fertilizers by locally-sourced substitutes, such as manure or compost, or a composite of local industrial phosphate mixed with local organic fertilizers) or decreasing quantities of inputs used, cultivating smaller areas (FAO, 2020). However, this overview remains general and somewhat limited for several reasons. First, the intensity of the crisis varies significantly across regions, leading to different local dynamics. Second, agricultural environments differ among farmers, influencing their capacity to adapt and make production choices. Most importantly, crop yield has not been estimated, even though it is a key factor since productivity directly affects the profit farmers can obtain. Therefore, while this analysis offers valuable insights, it remains a preliminary step. 5. Conclusion and Perspectives Our study successfully mapped winter wheat areas across Lebanon over successive seasons, from the 2017–2018 season up to 2024–2025, using remote sensing technology, and analyzed the inter-annual variability of these areas in the context of the country's multiple, overlapping crises. The outcomes of this research are both technical and agricultural. From a technical perspective, the RF classifier, used in our study, demonstrated strong performance and temporal transferability across the time series. The model trained on 2019–2020 data exhibited the highest consistency and was chosen for classification across the whole time period (2017–2018 to 2024–2025). Overall Accuracy for the estimation reached 87%. Wheat classification specifically showed consistently high Producer Accuracy (PA) and User Accuracy (UA) for most years, although a slight decline was observed in 2022–2023. The results confirm that classification maps alone can be biased if relying solely on pixel counts for area estimation. The statistically rigorous Random Stratified Sample (RSS) approach was applied to derive unbiased estimates and quantify classification uncertainty. The comparison between methods showed that the mapped wheat area was underestimated by pixel counts in all seasons except 2018–2019, with the bias reaching up to 4,272 ha. Therefore, incorporating reference samples is necessary for more accurate area estimation. From agricultural and socio-economical perspective, our study successfully monitored the fluctuation of wheat area, indicating that farmers’ decisions regarding wheat cultivation were significantly influenced by multiple factors related to the multi-factorial crisis threatening Lebanon. Wheat area generally increased until 2020–2021, but a sharp decrease of about 52% occurred in the 2021–2022 season, followed by a significant recovery in 2022–2023, reaching the highest level observed during the study period. The area decreased again by approximately 34% in 2023–2024 and 38% in 2024–2025. The observed variability showed little correlation with precipitation, suggesting that non-climatic factors are dominant, especially since over 80% of wheat plots in Lebanon are supplementary irrigated. The sharp decline in 2021–2022 was driven by the severe financial crisis, which included the suspension of the government subsidy program (removing a secure marketing channel), soaring inflation, and a huge increase in production costs, particularly for fuel and fertilizers. This made wheat cultivation a losing scenario, particularly for small farmers. The subsequent increase in 2022–2023 coincided with a small degree of economic stabilization, including partial dollarization, which encouraged farmers to reinvest in wheat production. The decrease in 2023–2024 and 2024–2025 was linked to the renewed instability and insecurity caused by the Israeli offensive on Lebanon starting in October 2023, which impacted main wheat production areas and caused displacement. The high interannual variability is also consistent with common Lebanese agricultural practices, particularly crop rotation (e.g., wheat-potato rotation). Using the available data, we attempted to explain this variability, though further research is needed to fully understand it. While we analyzed the macro-economic factors across Lebanon along with precipitation trends, in a way to gain a general understanding of this variability, and to establish a baseline understanding of the situation. In order, to explain this variability in a holistic way, detailed information on the agriculture productivity and the agricultural practices of wheat farmers and their decision-making processes, across the different regions in Lebanon, are required. This would help identify the factors that generally influence farmers’ crop choices, the changes that have occurred over time and their responses towards these changes; thereby, clarifying the reasons behind these patterns. Such information can be gathered through crop modeling for the productivity and socio-economic surveys with wheat farmers for the decision-making process. Consequently, future work should aim at deepening the analysis provided in this paper by addressing the wheat productivity over the cultivated areas and the practices of the farmers. Wheat productivity and yield will provide insights into the impact of fertilizers use, irrigation and pumping impacts, and access to fields during conflicts. Such exercise can be done by using the maps provided in this study as inputs to remote sensing driven crop modeling tools like Agri-Carbon- EO model (Al Bitar et al., 2022; Wijmer et al., 2024). In addition, agricultural practices and farmers decision-making will require adapted socio-economic surveys. Declarations Acknowledgment The authors gratefully acknowledge the financial support provided by the SAFAR program, which awarded a doctoral scholarship to Mariam Ibrahim; the PHC Cèdre program (project number 50473XL), which funded collaborative research activities between the French and Lebanese partners involved in this study; and the Water Diplomacy Center of Jordan (WDC) for providing a doctoral fellowship to Mariam Ibrahim. References Abay, K. A., Breisinger, C., Glauber, J., Kurdi, S., Laborde, D., & Siddig, K. (2023). The Russia-Ukraine war: Implications for global and regional food security and potential policy responses. Global Food Security , 36 , 100675. https://doi.org/10.1016/j.gfs.2023.100675 Abdallah, A. (2025, March 6). Meeting with Abbas Abdallah [Personal communication]. Abou Ltaif, S. F., Mihai-Yiannaki, S., Thrassou, A., Ltaif, S. F. A., Mihai-Yiannaki, S., & Thrassou, A. (2024). Lebanon’s Economic Development Risk: Global Factors and Local Realities of the Shadow Economy Amid Financial Crisis. Risks , 12 (8). https://doi.org/10.3390/risks12080122 Acevedo, E., Silva, P., & Silva, H. (2002). Wheat growth and physiology. In Bread Wheat Improvement and Production (pp. 39–70). FAO. Achli, S., Ongoma, V., Epule, T. E., Dhiba, D., Salih, W., Ousayd, L., & Chehbouni, A. (2025). Exploring the key drivers of crop yields in Morocco – a systematic review. Frontiers in Agronomy , 7 . https://doi.org/10.3389/fagro.2025.1515938 Aitkenhead, M. J., & Dyer, R. (2007). Improving Land-cover Classification Using Recognition Threshold Neural Networks. Photogrammetric Engineering & Remote Sensing , 73 (4), 413–421. https://doi.org/10.14358/PERS.73.4.413 Al Bitar, A., Wijmer, T., Arnaud, L., Fieuzal, R., J-F, S., Gibrin, H., Ferlicop, M., J-F, D., & Ceschia, E. (2022). Quantification of the impact of cover crops on Net Ecosystem Exchange using AgriCarbon-EOv0.1. IGARSS 2022–2022 IEEE International Geoscience and Remote Sensing Symposium , 5781–5784. https://doi.org/10.1109/IGARSS46834.2022.9883585 Alami Machichi, M., El Mansouri, L., Imani, Y., Bourja, O., Hadria, R., Lahlou, O., Benmansour, S., Zennayi, Y., & Bourzeix, F. (2022). CerealNet: A Hybrid Deep Learning Architecture for Cereal Crop Mapping Using Sentinel-2 Time-Series. Informatics , 9 (4), Article 4. https://doi.org/10.3390/informatics9040096 Amare, M., Abay, K. A., Berhane, G., Andam, K. S., & Adeyanju, D. (2025). Conflicts, crop choice, and agricultural investments: Empirical evidence from Nigeria. Land Use Policy , 148 , 107391. https://doi.org/10.1016/j.landusepol.2024.107391 Antonenko, V., Al Bitar, A., Danylenko, I., Wijmer, T., Colin, J., Dejoux, J.-F., Lefebvre, A., Knibbe, M., Ceschia, E., & Gascoin, S. (2024). Impact of the Russian invasion on wheat biomass in Ukraine. Environmental Research Letters . https://doi.org/10.1088/1748-9326/ad8363 Arafeh, N., & Sukarieh, M. (2023). Breaking the Cycle: Toward a New Imaginary of the Food System in Lebanon . https://carnegieendowment.org/research/2023/07/breaking-the-cycle-toward-a-new-imaginary-of-the-food-system-in-lebanon?center=middle-east&lang=en Bashour, I., Shehada Al-Ouda, A., Kassam, A., Bachour, R., Jouni, K., Hansmann, B., & Estephan, C. (2016). An overview of Conservation Agriculture in the dry Mediterranean environments with a special focus on Syria and Lebanon. AIMS Agriculture and Food , 1 , 67–84. https://doi.org/10.3934/agrfood.2016.1.67 Berio Fortini, L., Chen, Q., Uyehara, Y., Bogner, K., Sprague, J., & Sprague, R. (2024). Fine-resolution land cover mapping over large and mountainous areas for Lāna‘i, Hawaii using posterior probabilities, and expert knowledge. International Journal of Remote Sensing , 45 (6), 1949–1971. https://doi.org/10.1080/01431161.2024.2321465 Bhattarai, G. P., Schmid, R. B., & McCornack, B. P. (2019). Remote Sensing Data to Detect Hessian Fly Infestation in Commercial Wheat Fields. Scientific Reports , 9 (1), Article 1. https://doi.org/10.1038/s41598-019-42620-0 Boaretto, A. E., Spolidoirio, E. S., Trivelin, P. C. O., Muraoka, T., Freitas, J. G. de, & Cantarella, H. (2000). Optimizing nitrogen-fertilizer application to wheat under irrigation. In Optimizing nitrogen fertilizer application to irrigated wheat. Results of a co-ordinated research project. 1994–1998 (IAEA-TECDOC–1164; pp. 63–72). https://inis.iaea.org/records/rycfc-er588 Braga, P., Crusiol, L. G. T., Nanni, M. R., Caranhato, A. L. H., Fuhrmann, M. B., Nepomuceno, A. L., Neumaier, N., Farias, J. R. B., Koltun, A., Gonçalves, L. S. A., & Mertz-Henning, L. M. (2021). Vegetation indices and NIR-SWIR spectral bands as a phenotyping tool for water status determination in soybean. Precision Agriculture , 22 (1), 249–266. https://doi.org/10.1007/s11119-020-09740-4 Breiman, L. (1996). Bagging predictors. Machine Learning , 24 (2), 123–140. https://doi.org/10.1007/BF00058655 Breiman, L. (2001). Random Forests. Machine Learning , 45 (1), 5–32. https://doi.org/10.1023/A:1010933404324 Breisinger, C., Khouri, N., Glauber, J. W., & Laborde Debucquet, D. (2023). One of the world’s worst economic collapses, now compounded by the Ukraine crisis: What’s next for Lebanon? https://hdl.handle.net/10568/140097 Bwambale, E., Naangmenyele, Z., Iradukunda, P., Agboka, K. M., Houessou-Dossou, E. A. Y., Akansake, D. A., Bisa, M. E., Hamadou, A.-A. H., Hakizayezu, J., Onofua, O. E., & Chikabvumbwa, S. R. (2022). Towards precision irrigation management: A review of GIS, remote sensing and emerging technologies. Cogent Engineering , 9 (1), 2100573. https://doi.org/10.1080/23311916.2022.2100573 Caiserman, A. (2020). Adapter les stratégies agricoles aux évolutions socio-économiques et climatiques en milieu méditerranéen: Comparaison de l’usage de l’eau et des choix de plantes des agriculteurs dans la plaine de la Békaa (Liban) et de Marvdasht (Iran) [These de doctorat, Lyon]. https://www.theses.fr/2020LYSE3009 Caiserman, A., Dumas, D., Bennafla, K., Faour, G., & Amiraslani, F. (2019). Application of Remotely Sensed Imagery and Socioeconomic Surveys to Map Crop Choices in the Bekaa Valley (Lebanon). Agriculture , 9 (3), Article 3. https://doi.org/10.3390/agriculture9030057 Ceccato, P., Gobron, N., Flasse, S., Pinty, B., & Tarantola, S. (2002). Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1: Theoretical approach. Remote Sensing of Environment , 82 (2), 188–197. https://doi.org/10.1016/S0034-4257(02)00037-8 Chaves, M. E. D., Picoli, M. C. A., & Sanches, I. D. (2020). Recent Applications of Landsat 8/OLI and Sentinel-2/MSI for Land Use and Land Cover Mapping: A Systematic Review. Remote Sensing , 12 (18), Article 18. https://doi.org/10.3390/rs12183062 CNRS-L, & UNDP Lebanon. (2024). Israeli Offensive on Lebanon 2023–2024: Overview of attacks and damages across key sectors (p. 58). https://www.cnrs.edu.lb/Library/Files/Uploaded%20Files/CNRS-L%20report%20on%20Israeli%20Offensive%20against%20Lebanon%202023-2024%20English.pdf Congalton, R. G., & Green, K. (2019). Assessing the Accuracy of Remotely Sensed Data: Principles and Practices, Third Edition (3rd ed.). CRC Press. https://doi.org/10.1201/9780429052729 Dal, E., Díaz-González, A. M., Morales-Opazo, C., & Vigani, M. (2021). Agricultural sector review in Lebanon (FAO Agricultural Development Economics Technical Study No. 12). FAO; https://openknowledge.fao.org/handle/20. 500.14283/cb5157en Darwish, T., Zdruli, P., Saliba, R., Awad, M., Shaban, A., & Faour, G. (2012). Vulnerability to Desertification in Lebanon Based on Geo-information and Socioeconomic Conditions. Journal of Environmental Science and Engineering B . Delegido, J., Verrelst, J., Alonso, L., & Moreno, J. (2011). Evaluation of Sentinel-2 Red-Edge Bands for Empirical Estimation of Green LAI and Chlorophyll Content. Sensors , 11 (7), Article 7. https://doi.org/10.3390/s110707063 Devadoss, S., & Ridley, W. (2024). Impacts of the Russian invasion of Ukraine on the global wheat market. World Development , 173 , 106396. https://doi.org/10.1016/j.worlddev.2023.106396 European Space Agency. (2021). Copernicus Sentinel-2: MSI Level-2A BOA Reflectance Product. Collection 1 [Dataset]. https://doi.org/10.5270/S2_-znk9xsj FAO. (2006). Food Security [Policy Brief]. https://www.fao.org/fileadmin/templates/faoitaly/documents/pdf/pdf_Food_Security_Cocept_Note.pdf FAO. (2020). Special Report – FAO Mission to Assess the Impact of the Financial Crisis on Agriculture in the Republic of Lebanon . FAO. https://doi.org/10.4060/cb1164en FAO. (2023a). Food and Agriculture Organization Statistics . https://www.fao.org/faostat/en/#data/QCL FAO GIEWS. (2018, January 28). Global information and early warning system on food and agriculture (GIEWS) Country Brief: Lebanon 23-January-2018 . https://reliefweb.int/report/lebanon/giews-country-brief-lebanon-23-january-2018 FAO (with IFAD, UNICEF, WFP, & WHO). (2023b). The State of Food Security and Nutrition in the World 2023: Urbanization, agrifood systems transformation and healthy diets across the rural–urban continuum . FAO. https://doi.org/10.4060/cc3017en FAO, L. and W. D. (1971). A study on the response of wheat to fertilizers. (1st ed.). https://openknowledge.fao.org/handle/20.500 .14283/aq377e Faour, G., & Abdallah, C. (2020). Mapping Lands Cultivated with Wheat or Barley for the Year 2020 using Remote Sensing Technology (p. 234) [Internal Report]. CNRS Lebanon. Fayad, A., Gascoin, S., Faour, G., Fanise, P., Drapeau, L., Somma, J., Fadel, A., Al Bitar, A., & Escadafal, R. (2017). Snow observations in Mount Lebanon (2011–2016). Earth System Science Data , 9 (2), 573–587. https://doi.org/10.5194/essd-9-573-2017 Feller, W. (1991). An Introduction to Probability Theory and Its Applications, Volume 2 . John Wiley & Sons. Fernández-Manso, A., Fernández-Manso, O., & Quintano, C. (2016). SENTINEL-2A red-edge spectral indices suitability for discriminating burn severity. International Journal of Applied Earth Observation and Geoinformation , 50 , 170–175. https://doi.org/10.1016/j.jag.2016.03.005 Foody, G. M. (2002). Status of land cover classification accuracy assessment. Remote Sensing of Environment , 80 (1), 185–201. https://doi.org/10.1016/S0034-4257(01)00295-4 Gallego, J., Carfagna, E., & Baruth, B. (2010). Accuracy, Objectivity and Efficiency of Remote Sensing for Agricultural Statistics. In Agricultural Survey Methods (pp. 193–211). John Wiley & Sons, Ltd. https://doi.org/10.1002/9780470665480.ch12 Giri, C., Pengra, B., Long, J., & Loveland, T. R. (2013). Next generation of global land cover characterization, mapping, and monitoring. International Journal of Applied Earth Observation and Geoinformation , 25 , 30–37. https://doi.org/10.1016/j.jag.2013.03.005 Google for Developers. (2025). Google Earth Engine . https://developers.google.com/earth-engine Gordon, S. I. (1980). Utilizing LANDSAT imagery to monitor land-use change: A case study in ohio. Remote Sensing of Environment , 9 (3), 189–196. https://doi.org/10.1016/0034-4257(80)90028-0 Hall, D. K., Riggs, G. A., & Salomonson, V. V. (1995). Development of methods for mapping global snow cover using moderate resolution imaging spectroradiometer data. Remote Sensing of Environment , 54 , 127–140. https://doi.org/10.1016/0034-4257(95)00137-P Hashmiu, I., Agbenyega, O., & Dawoe, E. (2022). Determinants of crop choice decisions under risk: A case study on the revival of cocoa farming in the Forest-Savannah transition zone of Ghana. Land Use Policy , 114 , 105958. https://doi.org/10.1016/j.landusepol.2021.105958 Howarth, J. P., & Wickware, G. M. (1981). Procedures for change detection using Landsat digital data. International Journal of Remote Sensing , 2 (3), 277–291. Hunt, E. R., & Rock, B. N. (1989). Detection of changes in leaf water content using Near- and Middle-Infrared reflectances. Remote Sensing of Environment , 30 (1), 43–54. https://doi.org/10.1016/0034-4257(89)90046-1 Immitzer, M., Vuolo, F., & Atzberger, C. (2016). First Experience with Sentinel-2 Data for Crop and Tree Species Classifications in Central Europe. Remote Sensing , 8 (3), Article 3. https://doi.org/10.3390/rs8030166 IPCC. (2023). Summary for Policymakers. In Climate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change ([Core Writing Team, H. Lee and J. Romero (eds.)]., pp. 1–34). IPT Group. (2025). Fuel Prices . https://www.iptgroup.com.lb/Sub.aspx?pageid=22 Jaafar, H., Karimi, P., & Borgomeo, E. (2024). Economic irrigation water productivity of wheat and potato: An earth observation perspective on policy implications in the Litani Basin, Lebanon. Agricultural Water Management , 306 , 109180. https://doi.org/10.1016/j.agwat.2024.109180 Khabba, S., Er-Raki, S., Toumi, J., Ezzahar, J., Ait Hssaine, B., Le Page, M., & Chehbouni, A. (2020). A Simple Light-Use-Efficiency Model to Estimate Wheat Yield in the Semi-Arid Areas. Agronomy , 10 (10), Article 10. https://doi.org/10.3390/agronomy10101524 Khafagy, A., Díaz-González, A. M., Nano, E., Soguero Escuer, J., Morales Opazo, C., Salibi, A., El Tawn, L., & Dikah, W. (2022). Lebanon’s agrifood system in times of turbulence: Obstacles and opportunities . FAO. https://doi.org/10.4060/cc2334en Khan, A., Ahmad, M., Ahmed, M., & Iftikhar Hussain, M. (2020). Rising Atmospheric Temperature Impact on Wheat and Thermotolerance Strategies. Plants , 10 (1), 43. https://doi.org/10.3390/plants10010043 Khlif, M., Escorihuela, M. J., Chahbi Bellakanji, A., Paolini, G., Kassouk, Z., & Lili Chabaane, Z. (2023). Multi-Year Cereal Crop Classification Model in a Semi-Arid Region Using Sentinel-2 and Landsat 7–8 Data. Agriculture , 13 (8), Article 8. https://doi.org/10.3390/agriculture13081633 Koweyes, J., Salloum, T., Haidar, S., Merhi, G., & Tokajian, S. (2021). COVID-19 Pandemic in Lebanon: One Year Later, What Have We Learnt? mSystems , 6 (2), e00351-21. https://doi.org/10.1128/mSystems.00351-21 Kuhla, K., Puma, M. J., & Otto, C. (2024). International cooperation was key to stabilize wheat prices after the Russian Invasion of Ukraine. Communications Earth & Environment , 5 (1), 481. https://doi.org/10.1038/s43247-024-01638-7 Lago-Olveira, S., Ouhemi, H., Idrissi, O., Moreira, M. T., & González-García, S. (2024). Promoting more sustainable agriculture in the Moroccan drylands by shifting from conventional wheat monoculture to a rotation with chickpea and lentils. Cleaner Environmental Systems , 12 , 100169. https://doi.org/10.1016/j.cesys.2024.100169 Landis, J. R., & Koch, G. G. (1977). The Measurement of Observer Agreement for Categorical Data. Biometrics , 33 (1), 159–174. https://doi.org/10.2307/2529310 Le, T. H. L., Kristiansen, P., Vo, B., Moss, J., & Welch, M. (2024). Understanding factors influencing farmers’ crop choice and agricultural transformation in the Upper Vietnamese Mekong Delta. Agricultural Systems , 216 , 103899. https://doi.org/10.1016/j.agsy.2024.103899 Lebanese Customs. (2024). Trade Statistics . http://www.customs.gov.lb/Trade_Statistics/Yearly/Yearly_Statistics.aspx Lidon, F. C., Almeida, A. S., Leitao, A. L. cia, Silva, M. M., Pinheiro, N., Macas, B., & Costa, R. (2014). A SYNOPTIC OVERVIEW OF DURUM WHEAT PRODUCTION IN THE MEDITERRANEAN REGION AND PROCESSING FOLLOWING THE EUROPEAN UNION REQUIREMENTS. Emirates Journal of Food and Agriculture , 693–705. https://doi.org/10.9755/ejfa.v26i8.17066 Liu, J., Feng, Q., Gong, J., Zhou, J., Liang, J., & Li, Y. (2018). Winter wheat mapping using a random forest classifier combined with multi-temporal and multi-sensor data. International Journal of Digital Earth , 11 (8), 783–802. https://doi.org/10.1080/17538947.2017.1356388 Macintyre, P., van Niekerk, A., & Mucina, L. (2020). Efficacy of multi-season Sentinel-2 imagery for compositional vegetation classification. International Journal of Applied Earth Observation and Geoinformation , 85 , 101980. https://doi.org/10.1016/j.jag.2019.101980 Maddah, L., & Darwish, S. (2023). The Road to Recovery for Lebanese Agriculture . The Lebanese Center for Policy Studies. https://www.lcps-lebanon.org/en/articles/details/4821/the-road-to-recovery-for-lebanese-agriculture Main-Knorn, M., Pflug, B., Louis, J., Debaecker, V., Müller-Wilm, U., & Gascon, F. (2017). Sen2Cor for Sentinel-2. Image and Signal Processing for Remote Sensing XXIII , 10427 , 37–48. https://doi.org/10.1117/12.2278218 Makdisi, S., & Amine, R. (2022). Lebanon’s Multifaceted Economic Crisis of October 2019: Causes, Repercussions- A Diagnosis. Economic Research Forum , 1562 , Article 1562. https://ideas.repec.org//p/erg/wpaper/1562.html McFEETERS, S. K. (1996). The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. International Journal of Remote Sensing , 17 (7), 1425–1432. https://doi.org/10.1080/01431169608948714 Ministry of Agriculture, & FAO. (2010). الإحصاء الزراعي الشامل Comprehensive Agricultural Census . http://www.agriculture.gov.lb/Statistics-and-Studies/Comprehensive-Agricultural-Statistics Ministry of Finance. (2012). Lebanon—Wheat and Bread Subsidy Policies 2007–2011 . https://www.arabdevelopmentportal.com/publication/lebanon-wheat-and-bread-subsidy-policies-2007-2011 Myneni, R. B., & Williams, D. L. (1994). On the relationship between FAPAR and NDVI. Remote Sensing of Environment , 49 (3), 200–211. https://doi.org/10.1016/0034-4257(94)90016-7 Narin, O. G., Abdikan, S., Bayik, C., Sekertekin, A., Delen, A., & Balik Sanli, F. (2021). COHERENCE and BACKSCATTER BASED CROPLAND MAPPING USING MULTI-TEMPORAL SENTINEL-1 with DYNAMIC TIME WARPING. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences , XLIII-B5-2021 , 37–41. https://doi.org/10.5194/isprs-archives-XLIII-B5-2021-37-2021 NASA. (2023). Tiling System . https://hls.gsfc.nasa.gov/products-description/tiling-system/ Nasrallah, A. (2019). Crop mapping and yield estimation of wheat in the Bekaa plain of Lebanon [Phdthesis, AgroParisTech]. https://pastel.hal.science/tel-02520013 Nasrallah, A., Baghdadi, N., El Hajj, M., Darwish, T., Belhouchette, H., Faour, G., Darwich, S., & Mhawej, M. (2019). Sentinel-1 Data for Winter Wheat Phenology Monitoring and Mapping. Remote Sensing , 11 (19), Article 19. https://doi.org/10.3390/rs11192228 Nasrallah, A., Baghdadi, N., Mhawej, M., Faour, G., Darwish, T., Belhouchette, H., & Darwich, S. (2018). A Novel Approach for Mapping Wheat Areas Using High Resolution Sentinel-2 Images. Sensors (Basel, Switzerland) , 18 (7), 2089. https://doi.org/10.3390/s18072089 OCHA, R. (2024, December 23). Lebanon at War Crisis Update, November 2024 . https://reliefweb.int/report/lebanon/lebanon-war-crisis-update-november-2024 Oh, S., Chang, A., Ashapure, A., Jung, J., Dube, N., Maeda, M., Gonzalez, D., & Landivar, J. (2020). Plant Counting of Cotton from UAS Imagery Using Deep Learning-Based Object Detection Framework. Remote Sensing , 12 (18), Article 18. https://doi.org/10.3390/rs12182981 Olofsson, P., Arévalo, P., Espejo, A. B., Green, C., Lindquist, E., McRoberts, R. E., & Sanz, M. J. (2020). Mitigating the effects of omission errors on area and area change estimates. Remote Sensing of Environment , 236 , 111492. https://doi.org/10.1016/j.rse.2019.111492 Olofsson, P., Foody, G. M., Herold, M., Stehman, S. V., Woodcock, C. E., & Wulder, M. A. (2014). Good practices for estimating area and assessing accuracy of land change. Remote Sensing of Environment , 148 , 42–57. https://doi.org/10.1016/j.rse.2014.02.015 Olofsson, P., Foody, G. M., Stehman, S. V., & Woodcock, C. E. (2013). Making better use of accuracy data in land change studies: Estimating accuracy and area and quantifying uncertainty using stratified estimation. Remote Sensing of Environment , 129 , 122–131. https://doi.org/10.1016/j.rse.2012.10.031 Probst, P., & Boulesteix, A.-L. (2018). To Tune or Not to Tune the Number of Trees in Random Forest. Journal of Machine Learning Research , 18 (181), 1–18. Rodriguez-Galiano, V. F., Ghimire, B., Rogan, J., Chica-Olmo, M., & Rigol-Sanchez, J. P. (2012). An assessment of the effectiveness of a random forest classifier for land-cover classification. ISPRS Journal of Photogrammetry and Remote Sensing , 67 , 93–104. https://doi.org/10.1016/j.isprsjprs.2011.11.002 Rouse, J., Haas, R. H., Schell, J. A., & Deering, D. (1973). Monitoring vegetation systems in the great plains with ERTS . https://www.semanticscholar.org/paper/Monitoring-vegetation-systems-in-the-great-plains-Rouse-Haas/fb2f60fe0fe2874e 5cbf927a2556d719c32eac29 Safa, M., Samarasinghe, S., & Mohssen, M. (2010). Determination of fuel consumption and indirect factors affecting it in wheat production in Canterbury, New Zealand. Energy , 35 (12), 5400–5405. https://doi.org/10.1016/j.energy.2010.07.015 Sanlaville, P. (1963). Les régions agricoles du Liban . https://doi.org/10.3406/geoca.1963.1751 Science Toolbox Exploitation Platform, E. (2024). Sen2Cor. SNAP Supported Plugins . https://step.esa.int/main/snap-supported-plugins/sen2cor/ SINGH, A. (1989). Review Article Digital change detection techniques using remotely-sensed data. International Journal of Remote Sensing , 10 (6), 989–1003. https://doi.org/10.1080/01431168908903939 Sothe, C., Almeida, C. M. de, Liesenberg, V., & Schimalski, M. B. (2017). Evaluating Sentinel-2 and Landsat-8 Data to Map Sucessional Forest Stages in a Subtropical Forest in Southern Brazil. Remote Sensing , 9 (8), Article 8. https://doi.org/10.3390/rs9080838 Tadesse, W., Amri, A., Sanchez-Garcia, M., El-Bouhssini, M., Karrou, M., & Patil, S. (2017). Improving wheat production in the Central and West Asia and North Africa (CWANA) region. In Achieving sustainable cultivation of wheat (Vol. 2). Burleigh Dodds Science Publishing. Talhouk, S. N., Zurayk, R., & Khuri, S. (2001). Conservation of the coniferous forests of Lebanon: Past, present and future prospects. Oryx , 35 (3), 206–215. https://doi.org/10.1046/j.1365-3008.2001.00180.x Tawk, S. T., Chedid, M., Chalak, A., Karam, S., & Hamadeh, S. K. (2019). Challenges and Sustainability of Wheat Production in a Levantine Breadbasket: The Case of the West Bekaa, Lebanon. Journal of Agriculture, Food Systems, and Community Development , 8 (4), Article 4. https://doi.org/10.5304/jafscd.2019.084.011 Tita, D., Mahdi, K., Devkota, K. P., & Devkota, M. (2025). Climate change and agronomic management: Addressing wheat yield gaps and sustainability challenges in the Mediterranean and MENA regions. Agricultural Systems , 224 , 104242. https://doi.org/10.1016/j.agsy.2024.104242 Tohme Tawk, S., Chedid, M., Chalak, A., Karam, S., & Hamadeh, S. (2019). Challenges and Sustainability of Wheat Production in a Levantine Breadbasket: The Case of the West Bekaa, Lebanon. Journal of Agriculture, Food Systems, and Community Development , 1–17. https://doi.org/10.5304/jafscd.2019.084.011 UN Office for the Coordination of Humanitarian Affairs. (2020, August 7). Lebanon: Beirut Port Explosions Situation Report No. 2 (As of 7 August 2020) . https://reliefweb.int/report/lebanon/lebanon-beirut-port-explosions-situation-report-no-2-7-august-2020-enar United Nations. (2023). Lebanon Crisis Response Plan (LCRP) . https://data.unhcr.org/en/documents/details/100389 Verdeil, E., Faour, G., & Hamze, M. (2016). Atlas du Liban: Les nouveaux défis. In Atlas du Liban: Les nouveaux défis . Presses de l’Ifpo. https://books.openedition.org/ifpo/10709 Volsi, B., Higashi, G. E., Bordin, I., & Telles, T. S. (2022). The diversification of species in crop rotation increases the profitability of grain production systems. Scientific Reports , 12 (1), 19849. https://doi.org/10.1038/s41598-022-23718-4 Wagner, J., Skakun, S., Nair, S. S., Sadeh, Y., Baber, S., Oliinyk, O., Kotcharlakota,A., Gupta, M., Poliakov, D., Vaskivskyi, B., Misiura, O., Kussul, N., Prykhodko, D.,Sikachyna, O., Rajaoberison, A., Li, F., Chevassu, M., Rehbinder, J., Nerry, F., …Becker-Reshef, I. (2026). Monitoring winter crop areas during wartime: Remote sensing support for Ukraine’s agricultural statistics. Npj Sustainable Agriculture , 4 (1), 1. https://doi.org/10.1038/s44264-025-00119-4 Wang, D., Li, Z., Zhou, Q., Chen, Z., & Liu, J. (2015). Estimation of winter wheat acreage via a combination of remotely sensed data and an optimized spatial sampling scheme. International Journal of Remote Sensing , 36 (19–20), 5208–5221. https://doi.org/10.1080/01431161.2015.1093197 Wijmer, T., Al Bitar, A., Arnaud, L., Fieuzal, R., & Ceschia, E. (2024). AgriCarbon-EO v1.0.1: Large-scale and high-resolution simulation of carbon fluxes by assimilation of Sentinel-2 and Landsat-8 reflectances using a Bayesian approach. Geoscientific Model Development , 17 (3), 997–1021. https://doi.org/10.5194/gmd-17-997-2024 World Bank Group. (2022). Lebanon—Wheat Supply Emergency Response Project . http://documents.worldbank.org/curated/en/408131653327258940 Yang, X., Xiong, J., Du, T., Ju, X., Gan, Y., Li, S., Xia, L., Shen, Y., Pacenka, S., Steenhuis, T. S., Siddique, K. H. M., Kang, S., & Butterbach-Bahl, K. (2024). Diversifying crop rotation increases food production, reduces net greenhouse gas emissions and improves soil health. Nature Communications , 15 (1), Article 1. https://doi.org/10.1038/s41467-023-44464-9 Zhao, Z., Islam, F., Waseem, L. A., Tariq, A., Nawaz, M., Islam, I. U., Bibi, T., Rehman, N. U., Ahmad, W., Aslam, R. W., Raza, D., & Hatamleh, W. A. (2024). Comparison of Three Machine Learning Algorithms Using Google Earth Engine for Land Use Land Cover Classification. Rangeland Ecology & Management , 92 , 129–137. https://doi.org/10.1016/j.rama.2023.10.007 Table 10 Table 10 is not available with this version. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 07 May, 2026 Reviewers agreed at journal 08 Apr, 2026 Reviews received at journal 05 Apr, 2026 Reviewers agreed at journal 03 Apr, 2026 Reviewers agreed at journal 08 Mar, 2026 Reviewers agreed at journal 13 Feb, 2026 Reviewers invited by journal 26 Jan, 2026 Editor assigned by journal 26 Jan, 2026 Submission checks completed at journal 23 Jan, 2026 First submitted to journal 20 Jan, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8649264","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":580798087,"identity":"ebfb7c69-c52c-4ab7-b195-b45f4f506225","order_by":0,"name":"Mariam Ibrahim","email":"","orcid":"","institution":"National Center for Remote Sensing (NCRS), National Council for Scientific Research of Lebanon (CNRS-L)","correspondingAuthor":false,"prefix":"","firstName":"Mariam","middleName":"","lastName":"Ibrahim","suffix":""},{"id":580798092,"identity":"d97d2aa7-2ab8-4c7b-8135-7d60918c5245","order_by":1,"name":"Ghaleb Faour","email":"","orcid":"","institution":"National Center for Remote Sensing (NCRS), National Council for Scientific Research of Lebanon (CNRS-L)","correspondingAuthor":false,"prefix":"","firstName":"Ghaleb","middleName":"","lastName":"Faour","suffix":""},{"id":580798093,"identity":"7c8b6fba-9e26-40fd-862e-79bee3ae0e62","order_by":2,"name":"Michel Le Page","email":"","orcid":"","institution":"Centre d'Études Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNES/CNRS/INRAE/IRD","correspondingAuthor":false,"prefix":"","firstName":"Michel","middleName":"Le","lastName":"Page","suffix":""},{"id":580798094,"identity":"33cc7fc8-16f8-4218-b9c1-b6176f26cefc","order_by":3,"name":"Marielle Montginoul","email":"","orcid":"","institution":"UMR G-Eau, Université de Montpellier, INRAE","correspondingAuthor":false,"prefix":"","firstName":"Marielle","middleName":"","lastName":"Montginoul","suffix":""},{"id":580798095,"identity":"3065cdc3-37f5-48f3-a419-f8ac8c435c0b","order_by":4,"name":"Ahmad Al Bitar","email":"","orcid":"","institution":"Centre d'Études Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNES/CNRS/INRAE/IRD","correspondingAuthor":false,"prefix":"","firstName":"Ahmad","middleName":"Al","lastName":"Bitar","suffix":""},{"id":580798096,"identity":"36b6da7f-e584-4cd6-9d70-2f67774f7f66","order_by":5,"name":"Bilal Komati","email":"","orcid":"","institution":"National Center for Remote Sensing (NCRS), National Council for Scientific Research of Lebanon (CNRS-L)","correspondingAuthor":false,"prefix":"","firstName":"Bilal","middleName":"","lastName":"Komati","suffix":""},{"id":580798098,"identity":"f315c649-1418-40e6-ad8f-905a9917ffe1","order_by":6,"name":"Lionel Jarlan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIie3QPQrCMBTA8Rce1KUfa6SgV3guoijeRQQd1c1BpFLo5AFaFI/hXCnUJXgBF48guAg6mKjglnYUzH9IMuQHLwEwmX4wF9U6AblhADDjACk7a4n1IqQIk0QogqQn8CbwIiyCEqSC+flCUPHW++V1um2N3RSti34wa9SI5WD82A/9ZMfbSYoY64nd9G31FsEC39lxotTL9IMp8pCkLlh4dzaKIBYT9WMkWOQ7QSliDasr4tiQpGPnnKpZAfG8LOe3WXdQE5id7PmC3EOoJ5/44HsuBWS9kvdMJpPpH3sCKw81rU6PCikAAAAASUVORK5CYII=","orcid":"","institution":"Centre d'Études Spatiales de la Biosphère (CESBIO), Université de Toulouse, CNES/CNRS/INRAE/IRD","correspondingAuthor":true,"prefix":"","firstName":"Lionel","middleName":"","lastName":"Jarlan","suffix":""}],"badges":[],"createdAt":"2026-01-20 12:42:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8649264/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8649264/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101785140,"identity":"8453008b-f891-4d6c-9c5d-3655badd81ea","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":333967,"visible":true,"origin":"","legend":"\u003cp\u003eEvolution of wheat supply from local production and imports in Lebanon 2018-2023 (FAOSTAT, 2025).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/0b66ddd9dd91b5ed32b10d30.png"},{"id":101785149,"identity":"15eebf06-3b08-4cbc-b0ac-93d3e5ddd801","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":3188820,"visible":true,"origin":"","legend":"\u003cp\u003eMaps showing Lebanon’s location and the main agricultural regions in the North, the Bekaa Valley, and the South. The 2017 LULC map is adapted from the National Center for Remote Sensing (CNRS-L). Sentinel-2 images from June 2017, downloaded from the Copernicus Browser.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/32156c2535e176b1bcf3953b.png"},{"id":101785141,"identity":"e45e49a8-95ee-4e40-8fc1-d6382494628f","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":481671,"visible":true,"origin":"","legend":"\u003cp\u003eTrends in seeds, fertilizer and fuel prices in Lebanon.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/0263629eab43b6edbb3f056f.png"},{"id":101785142,"identity":"01a8d9e3-dfd4-4267-a042-5f2ff9d6076a","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1563329,"visible":true,"origin":"","legend":"\u003cp\u003eDistribution of reference points over Lebanon\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/a04a3e7b405ee6c2d9399177.png"},{"id":101880645,"identity":"8fb55aff-b55a-40e6-969a-17cc23006d96","added_by":"auto","created_at":"2026-02-04 15:04:37","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":681930,"visible":true,"origin":"","legend":"\u003cp\u003eClassification workflow.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/60ca66834063fbabc09778f6.png"},{"id":101881398,"identity":"127dac81-7fc7-43d3-a52b-d4846d82a907","added_by":"auto","created_at":"2026-02-04 15:11:53","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":782797,"visible":true,"origin":"","legend":"\u003cp\u003ePost-classification workflow.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/313b40c06defbf71a1c244f2.png"},{"id":101880482,"identity":"3cf4df4b-4607-4bb2-9197-dea9c97aca77","added_by":"auto","created_at":"2026-02-04 15:02:40","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":1131932,"visible":true,"origin":"","legend":"\u003cp\u003eTransition matrix (percentage) by class over the study period. (2018→2019 means the transition from season 2017-2018 to 2018-2019).\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/bb757353d37fcad48be79e6d.png"},{"id":101880960,"identity":"489bd415-5b51-4451-a02e-f039e9ae3e89","added_by":"auto","created_at":"2026-02-04 15:08:21","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":2816414,"visible":true,"origin":"","legend":"\u003cp\u003eWheat pixels in a selected region of the Bekaa for seven cropping seasons: (A) 2017-2018, (B) 2018-2019, (C) 2019-2020, (D) 2020-2021, (E) 2021-2022, (F) 2022-2023, (G) 2023-2024 and (H) 2024-2025. (Sentinel-2 image on 30-04-2020 downloaded from Copernicus Browser)\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/fe47ae60166405aba43fd609.png"},{"id":101785152,"identity":"8865866f-39ac-4e9e-aef0-ebf2d33f9ac8","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":308708,"visible":true,"origin":"","legend":"\u003cp\u003eWheat area dynamics in relation to seasonal precipitation.\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/a1b170dc6f67d987d6665e54.png"},{"id":101880814,"identity":"22f72752-dc70-48d6-b3ac-e92d25e3afe9","added_by":"auto","created_at":"2026-02-04 15:06:39","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":375277,"visible":true,"origin":"","legend":"\u003cp\u003eChange in wheat pixel counts based on confidence and score margin combinations; error bars represent the interquartile range.\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/2469f6016d4cdfb83202775d.png"},{"id":101785150,"identity":"267bff1d-2e27-440d-acd9-5eed5c4a1035","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":557746,"visible":true,"origin":"","legend":"\u003cp\u003eWheat area trends under crisis pressure in Lebanon (2018-2025)\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/b709f5e0564108572e850b4a.png"},{"id":101785146,"identity":"8459f74f-449b-42b3-96a9-677e51dcd028","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":483433,"visible":true,"origin":"","legend":"\u003cp\u003eTotal wheat production cost (LBP/ha) and (bread bundle equivalent/ha) with exchange rate and food price inflation percentage.\u003c/p\u003e","description":"","filename":"12.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/83ebe997ff281ce23f7280c9.png"},{"id":101785148,"identity":"a7392cd3-8966-408f-b500-2022e48c501c","added_by":"auto","created_at":"2026-02-03 15:33:58","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":411179,"visible":true,"origin":"","legend":"\u003cp\u003eWheat loss percentage in the three main wheat-growing regions of Lebanon.\u003c/p\u003e","description":"","filename":"13.png","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/4d14f50edfdb7d1132049310.png"},{"id":101943150,"identity":"0d761d38-312c-4f46-a31b-a277d6d254a5","added_by":"auto","created_at":"2026-02-05 09:40:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":18187190,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8649264/v1/8a27a729-926a-46c2-9297-c8bf84b83f8e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Wheat in Crisis: Variability of Wheat Area in Lebanon 2017-2025","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eWheat is the most critical staple crop in the global food system, and one of the most politically significant commodities in the world. Wheat domestication was founded in the Fertile Crescent of the Near East, near the Tigris and Euphrates rivers, specifically in what is now southeastern Turkey and northern Syria, where its wild progenitor was found (Lidon et al., 2014). Today, wheat is cultivated across the globe, covering over 220\u0026nbsp;million hectares worldwide. It is also the top produced cereal in Lebanon according to the Food and Agriculture Organization (FAO) statistics (FAO, 2023a).\u003c/p\u003e \u003cp\u003eLocal production covers about 20% of Lebanon\u0026rsquo;s total wheat demand, while the 80% imports are provided by Ukraine (80%) and Russia (16%) (FAO, 2023a; World Bank Group, 2022) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Relying on imports means a high sensitivity to any external disruptions, leading to a lack of resilience and stability. Therefore, in times of crisis, the situation can be concerning.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIndeed, Lebanon represents a particularly acute case study of agricultural vulnerability under compounding crises. The economic breakdown began in October 2019 (Makdisi \u0026amp; Amine, 2022), and was marked by currency devaluation, hyperinflation, banking sector failure, and the deterioration of essential infrastructure and services. Lebanon was hit hard by the COVID-19 pandemic soon after (Koweyes et al., 2021). Just as the country was struggling to recover, the devastating Beirut port explosion in 2020 destroyed key grain silos (UN Office for the Coordination of Humanitarian Affairs, 2020). Then, the conflict between Russia and Ukraine, the two major wheat import sources, disrupted global wheat markets (Abay et al., 2023; Devadoss \u0026amp; Ridley, 2024). Additionally, since October 2023, the Israeli attacks on Lebanon hit main wheat production areas (CNRS-L \u0026amp; UNDP Lebanon, 2024). All of this unfolded under the looming impact of climate change (IPCC, 2023). These intersecting crises produced a multifaceted challenge for the Lebanese agricultural sector, and in particular wheat production. Food security in Lebanon has deteriorated dramatically, with the country facing unprecedented levels of food insecurity across multiple dimensions: availability, access, utilization, and stability (FAO, 2006).\u003c/p\u003e \u003cp\u003eAgainst this backdrop of crisis, monitoring wheat production becomes particularly relevant for understanding the challenges and opportunities within the Lebanese agricultural sector (FAO, 2023b). By accurately tracking the spatial and temporal dynamics of wheat cultivation, such a system provides the foundation to link production patterns with input costs, macroeconomic crisis indicators, and strategic food security considerations. Linking wheat production outcomes to these factors, enables valuable insights into how well the local production systems are performing and where adjustments may be needed and allows policymakers and stakeholders to design and implement more targeted and effective interventions, to enhance domestic food production capacity, optimize resource allocation and strengthen farmers\u0026rsquo; resilience under challenging conditions. Unfortunately, the Lebanese agricultural sector faces significant challenges, complicating wheat production monitoring. Opportunistic farming and fragmented land ownership, with 70% of farmers cultivating less than one hectare, lead to unpredictable production patterns and diverse farming practices and field conditions (Caiserman et al., 2019; Dal et al., 2021). The lack of continuous and up-to-date data on wheat areas, with the last comprehensive agricultural census conducted in 2010, further undermines any baseline for effective production monitoring (Ministry of Agriculture \u0026amp; FAO, 2010; Wang et al., 2015).\u003c/p\u003e \u003cp\u003eWhile these challenges are significant, they do not preclude the crucial need for monitoring system. The traditional crop fields identification survey, while offering high accuracy, present significant limitations including high costs, time-consuming fieldwork, and limited accessibility to certain areas. Thus, in practice, mapping becomes a compromise between the ideal and the feasible. Remote sensing emerged as transformative solution for these constraints, offering spatially and temporally cost-effective data for land-cover and land-use information derivation. Whether performing a simple plant count (Oh et al., 2020) or going further to estimate yield (Khabba et al., 2020), detect diseases (Bhattarai et al., 2019), manage irrigation (Bwambale et al., 2022), or monitor agricultural areas during wartime (Wagner et al., 2026), remote sensing technologies have evolved, particularly with the availability of Sentinel-2 (S2) imagery with 10-20m spatial resolution (European Space Agency, 2021), a substantial improvement over earlier coarse resolution (300m \u0026minus;\u0026thinsp;1Km) systems which often result in mixed pixels that misrepresent land cover, especially for small and locally managed land parcels (Giri et al., 2013). The integration of readily accessible cloud-computing platforms like Google Earth Engine (GEE) which offers extensive data catalog and tools for application development (Google for Developers, 2025), combined with the advancements in machine learning and deep learning algorithms, has potentially improved the process. Random Forest (RF) classifier has demonstrated strong performance in classification and is considered more user friendly, compared to commonly used classifiers, including Maximum Likelihood (ML), Support Vector Machine (SVM), Artificial Neural Network (ANN) (Liu et al., 2018; Zhao et al., 2024), and temporal convolutional neural network (Antonenko et al., 2024).\u003c/p\u003e \u003cp\u003eResearch across the Middle East and North Africa (MENA) region, demonstrates the effectiveness of these approaches. In Tunisia, combining Sentinel-2 images with 15 vegetation indices and RF algorithm achieved an overall accuracy ranging from 85.8% to 95.1% (Khlif et al., 2023). Morocco\u0026rsquo;s CerealNet, a hybrid deep learning architecture, combining a Long Short-Term Memory (LSTM) branch for phenological features extraction from the Normalized Difference Vegetation Index (NDVI) time series with a Convolutional Neural Network (CNN) branch for spectral features analysis, achieved a categorical accuracy of 95% (Alami Machichi et al., 2022). In Turkey, Sentinel-1 data, using coherence data alone yielded the highest accuracy of around 81% in crop classification (Narin et al., 2021). In Lebanon, two NDVI-based methodologies were applied specifically to the Bekaa plain: the Simple and Effective Wheat Mapping Approach (SEWMA), which exploits wheat's unique phenological patterns to distinguish it from barley and triticale, achieved 82.6\u0026ndash;87% accuracy (Nasrallah et al., 2018) while Euclidean distance classification of NDVI profiles from 386 ground-referenced fields reached 84.06% accuracy for wheat identification (Caiserman et al., 2019). While these findings demonstrate the effectiveness of remote sensing in agricultural monitoring, and particularly, in Lebanon, yet, these applications were conducted prior to the multi-dimensional crisis and did not account for the combined effects of economic, political, and socio-environmental shocks on wheat production, leaving a critical gap in understanding production dynamics under extreme conditions.\u003c/p\u003e \u003cp\u003eBuilding on this gap, our main objective is to study the variability of wheat area in the context of the multiple crises that occurred in Lebanon. In the subsequent sections, the study area and data together with the methodological materials for the classification and the assessment of uncertainties are described (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Followed by the results section (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) which is organized into three subsections on the classification accuracy of the maps and the associated uncertainties, as well as the interannual variability of wheat area. The discussion section (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) is also organized into two subsections where we discuss the wheat mapping approach and the link with the multi-crises in Lebanon. The conclusion section (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e) summarizes the main findings of the paper and opens into future investigations.\u003c/p\u003e"},{"header":"2. Material and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Geographical Description and Agricultural Background of the Study Area\u003c/h2\u003e \u003cp\u003eLebanon is situated on the eastern shore of the Mediterranean Sea, positioned between latitudes 33\u0026deg; and 35\u0026deg; N and longitudes 35\u0026deg; and 37\u0026deg; E (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Covering an area of 10,452 Km\u0026sup2;, Lebanon is predominantly mountainous with alternating geographical features. Landforms, climate, soils, and vegetation can vary significantly over short distances. The country is defined by four main geomorphological units: a narrow coastal plain, two mountain ranges (Mount Lebanon and the Anti-Lebanon) considered as the water tower of the country, and a fertile depression known as the Bekaa Valley, which lies at an altitude of 700 to 1,100 meters. In addition, there is the northern Akkar plains which are of interest for this study. Lebanon experiences a Mediterranean climate (Csa according to the K\u0026ouml;ppen-Geiger classification) characterized by four distinct seasons: a hot and dry summer, a cool and rainy winter, and moderately dry fall and spring seasons.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe coastal and mountainous areas receive abundant rainfall, up to 850 and 1,100 mm respectively, mostly between December and March. In contrast, the Bekaa Valley experiences a semi-arid to continental climate with unpredictable rainfall. Precipitation decreases from south to north, and its irregularity increases along this gradient (Sanlaville, 1963). The southern Bekaa Valley, has a sub-humid Mediterranean climate for example, and receives 500 to 800 mm of precipitation annually, while the north-eastern part is almost arid to continental, as it is shielded from the sea\u0026rsquo;s influence by a high mountain ridge exceeding 3,000 m in altitude and gets less than 600 mm. At very high elevations, snow patches can persist for 6\u0026ndash;9 months. In contrast, periods of total drought can occur, making complementary irrigation necessary for any form of intensive cultivation (Darwish et al., 2012; Fayad et al., 2017; Verdeil et al., 2016).\u003c/p\u003e \u003cp\u003eAgriculture plays an important role in Lebanon. Spanning over 200,000 ha, the country has a relatively diversified agricultural land, including fruits, vegetables, cereals, grapes, and olives (Dal et al., 2021; Maddah \u0026amp; Darwish, 2023). Lebanon has three main wheat-growing regions: the north, the south, and the Bekaa Plain. Among these, the Bekaa Plain is the largest contributor (Caiserman et al., 2019), accounting for an average of 58% of the total harvested area. Indeed, the Bekaa Plain offers the ideal environment for wheat cultivation and is often called Lebanon\u0026rsquo;s breadbasket. Stretching between the slopes of the Lebanon and Anti-Lebanon mountains, the Bekaa Plain features fertile alluvial soil, providing a rich nutrient base that supports wheat growth. With cold winters and hot summers, the crop benefits from low temperatures during the vernalization phase (Acevedo et al., 2002) and higher temperatures during heading, anthesis, and grain filling (Khan et al., 2020).\u003c/p\u003e \u003cp\u003eIn general, winter wheat is sown in autumn, enters dormancy in winter, begins its green-up phase in early spring, grows rapidly until the heading phase in late spring and reaches full maturity in early summer (Caiserman, 2020; Nasrallah, 2019). It\u0026rsquo;s hard to define generalized wheat cultivation practices, as it depends on each region and each farmer means. However, in general, the season begins with plowing the land, followed by sowing the grains, and then leveling the soil (A. Abdallah, personal communication, March 6, 2025).\u003c/p\u003e \u003cp\u003eThe farmer usually fertilizes in two stages: in February, nitrogen fertilization through ammonium fertilizer is used, and in April, urea is applied. Depending on his means, the farmer may use only urea or only ammonium. Very few can afford to use potash fertilizers or 20-20-20 (20% nitrogen, 20% phosphorus, and 20% potassium) during irrigation. Some farmers avoid using synthetic fertilizers, mainly because potatoes, often grown as the preceding crop (generally cultivated from early spring to late summer of the previous year) leave behind sufficient soil fertility to meet wheat\u0026rsquo;s nitrogen requirements (Nasrallah, 2019).\u003c/p\u003e \u003cp\u003eDespite being a winter crop, wheat often receives complementary irrigation in early spring to improve yields. In the Bekaa plain, over 80% of wheat plots are complementary irrigated, with the frequency depending directly on the rainfall received during the season. Since rainfall usually stops in February-March, farmers typically begin irrigating in mid-April and continue until mid-May. Some wheat plots, however, remain rainfed due to lack of access to water, financial constraints, or reliance on the hope of a good rainfall season (Nasrallah, 2019).\u003c/p\u003e \u003cp\u003eRegarding weed control, 2,4-D is used, and only a few uses a product specifically targeting wild oat. Concerning rust, which may affect wheat due to humidity, azoxystrobin, penconazole, or fosetyl-aluminum may be sprayed, once rust is observed (A. Abdallah, personal communication, March 6, 2025).\u003c/p\u003e \u003cp\u003eFarmers usually follow crop rotation seasons that include wheat, potato, corn and vegetables meaning that crop types in the agricultural areas are highly changing from one year to another. In the Bekaa valley, wheat-potato rotation is the most followed, compared to other type of rotations (Nasrallah, 2019).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Agro-Economic Context and Macro-Economic Data\u003c/h2\u003e \u003cp\u003eThe timeline of our study coincides with a cascade of overlapping and sequential crises that together have placed enormous stress on the agriculture sector since 2019. These multiple crises impacted the production costs and limited the field access for the farmer. This complex situation has led to direct macroeconomic impacts, most notably rising prices, and, consequently, to a gradual adaptation by farmers that has profoundly reshaped cereal cultivation in the country. A previous study has shown that wheat was chosen primarily for its profitability, with most farmers selling their production to the government, as part of the subsidy program (Caiserman et al., 2019). The economic crisis alone, led to (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) the suspension of the subsidy program, (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) a severe currency depreciation, (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) banking restrictions reduced access to credit for farmers, (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) a rise in prices of fuel, seeding and fertilizers due to local inflation and to a lesser extent to the global market. These multiple crisis-related impacts led to a drastic increase in production costs, which in turn probably resulted in a decline in cultivated areas as anticipated by FAO (2020). For instance, nitrogenous fertilizers prices peak in 2022. Within this context, many producers consequently reduced input use or shifted to low-input crops (Khafagy et al., 2022). The 2021 removal of fuel subsidies further raised irrigation, transport and mechanical tillage costs while border hostilities in southern Lebanon since late 2023 have displaced rural populations. The covid crisis limits access to fields for farmers. The 2020 port explosion compounded pre-existing supply-chain and currency constraints affecting imported inputs such as seeds. FAO reports indicate that seeds and other imported inputs remained available but at elevated prices in late 2020. The consequences for wheat cultivation are still poorly documented. Some studies highlighted that smallholders often abandoned cereal cultivation and migrated to cities in search of income (Arafeh \u0026amp; Sukarieh, 2023). Medium-sized farms reduced fertilizer and pesticide applications or switched to rainfed barley, legumes, or fallow systems. Larger farms adjusted cropping patterns or invested in limited mechanization consolidation. By lack of information on farmer adaptation strategy, quantitative assessment of the impact of crisis on cereal cropping area will focus on the macro-economic consequences for the farmers and, in particular, to the production costs.\u003c/p\u003e \u003cp\u003eFuel prices was collected from IPT Group (IPT Group, 2025), while average fertilizer prices were sourced from the Lebanese Customs website (Lebanese Customs, 2024). Seeds prices were provided by Maison de d\u0026rsquo;Agriculteur (A. Abdallah, personal communication, March 6, 2025). Fertilizer and seed prices are provided in US dollar per ton while fuel is in Lebanese pound per 20 liters. All data correspond to an average at the country scale and are provided on a yearly basis due to the lack of a complete infra-annual price series (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFrom these datasets, a simplified indicator of crop production cost per hectare of wheat accounting for fertilizers, seeds and fuel for mechanical tillage operations was computed based on representative values for wheat crop (per hectare) of 120 Kg of seeds (Bashour et al., 2016 for Mediterranean wheat systems), 80 Kg of nitrogen (Boaretto et al., 2000; FAO, 1971) and ~\u0026thinsp;65 L diesel.ha⁻\u0026sup1; (order-of-magnitude operational fuel use for wheat production following Safa et al., 2010). To best reflect the local conditions faced by farmers in Lebanon, all prices were also converted into Lebanese pounds using an annual average parallel-market exchange rate compiled from multiple online sources: 1,507 LBP/USD for the pre-crisis period (until 2019), 8,900 LBP/USD in 2020, 20,000 in 2021, 35,000 in 2022, 110,000 in 2023 and 90,000 in 2024 and 2025 (in the absence of values for 2025, we assumed a stable average exchange rate reflecting the post-crisis dollarized environment). Finally, we also expressed the production costs in bread-equivalent units to make the numbers easier to interpret by relating them to the price of a basic staple. To this objective, the average annual prices of a bread bundle were extracted from the Lebanon Market Monitor report produced by the World Food Program (WFP) accessible at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://dataviz.vam.wfp.org/the-middle-east-and-northern-africa/lebanon/reports?current_page=1\u0026amp;country=lbn\u003c/span\u003e\u003cspan address=\"https://dataviz.vam.wfp.org/the-middle-east-and-northern-africa/lebanon/reports?current_page=1\u0026amp;country=lbn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Ground Truth and Reference Points\u003c/h2\u003e \u003cp\u003eAs the focus is put on wheat crop areas, the classification scheme was intentionally limited to five major land-cover classes: wheat, bare soil, built-up, forest (including natural forest and agroforestry fields), and vegetation gathering all type of annuals, excluding wheat). This simplification reflects both the heterogeneous nature of Lebanese agricultural landscapes and the study\u0026rsquo;s single-crop focus. Previous research has shown that limiting the number of non-target vegetation classes enhances separability and classification accuracy in phenology-based crop mapping (Immitzer et al., 2016; Liu et al., 2018; Nasrallah et al., 2018). Moreover, in a crisis context with restricted field data availability, a reduced number of classes provides a balanced training dataset and greater Random Forest stability (Probst \u0026amp; Boulesteix, 2018). These five categories represent the dominant land types across Lebanon during the wheat season and account for nearly all spectral variability relevant to this analysis. Water surfaces and snow were excluded from the classification process (see section \u003cb\u003e2.4 below\u003c/b\u003e). Reference datasets consisted of two groups: wheat reference points gathered from ground surveys, and user defined reference points for the remaining classes, collected using high-resolution images from Google Earth.\u003c/p\u003e \u003cp\u003eWith regards to wheat ground truth, the Ministry of Economy and the Ministry of Agriculture have developed economic policies to support wheat farming and to encourage its continued cultivation until the season 2019\u0026ndash;2020. The government purchased wheat from farmers, with a payment ceiling typically based on the amount of wheat produced per unit of cultivated land (Ministry of Finance, 2012). The National Council for Scientific Research of Lebanon (CNRS-L) was officially responsible for this process, with the primary goal of accurately identifying areas that applied for subsidies.\u003c/p\u003e \u003cp\u003eWheat reference points were acquired from the CNRS-L database. For the crop seasons from 2017\u0026ndash;2018 up to 2019\u0026ndash;2020, CNRS-L digitized wheat property requests. Teams of engineers conducted field surveys across key regions to cover all plots (Faour \u0026amp; Abdallah, 2020). The remaining additional reference data was acquired in the 2022\u0026ndash;2023 season. For the latter, wheat reference points were provided by the Directorate General of Cereals and Beetroot (DGCB) - Ministry of Economy and Trade, based on a field survey they conducted that year. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the total number of fields surveyed (including wheat and other crops). Wheat fields were extracted from CNRS-L and DGCB databases and cross-checked using Google Earth. Then, reference points were systematically selected within the mapped polygons to ensure precise delineation and assessment of wheat coverage.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eNumber of surveyed fields by CNRS-L and DGCB teams.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWheat Season\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNumber of Surveyed Fields\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e2017\u0026ndash;2018\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3952\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e2018\u0026ndash;2019\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3740\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e2019\u0026ndash;2020\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e3038\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003e2022\u0026ndash;2023\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1427\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe reference points for other land cover classes were selected manually using high resolution images from Google Earth Pro. This method is considered reliable for validation data collection (Kennedy et al., 2007). Images covering the period from April to July of each of the four seasons (2017\u0026ndash;2018, 2018\u0026ndash;2019, 2019\u0026ndash;2020, and 2022\u0026ndash;2023) were used. Built-up, bare soil and forest points are easily detected by visual interpretation. To identify non-wheat points, and to prevent collecting points of non-subsidized wheat, we focused on the end of the growing season, when wheat will either be harvested or mature (yellow), meaning that high resolution google earth images were selected after May, 15. Areas with greenness at this stage can be considered as non-wheat (provided in the supplementary material \u003cb\u003eFigure S1\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eConsidering that the core objective of this study is to address the crisis scenario, one of the major implications of the crisis is the reduced accessibility to fields and logistical constraints, which directly affect the ability to collect reference data. For this reason, we intentionally limited the reference datasets to 550 points per season (110 points per class), as we also aimed to account for a potential data shortage that could emerge in the development of the monitoring system. It is important to note that, although a limited number of the available wheat fields were used to simulate a crisis context, all available wheat fields were ultimately employed to assess the classifier\u0026rsquo;s output (see \u003cb\u003esection 2.5.2\u003c/b\u003e).\u003c/p\u003e \u003cp\u003eThe four datasets (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) were used to train the classifier, and evaluate its accuracy. These datasets were also used as random stratified sample to estimate the wheat area accuracy, as described in \u003cb\u003esection 2.5.3\u003c/b\u003e (Olofsson et al., 2014). All the reference datasets were then uploaded to the GEE platform through a shape asset.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Satellite Data\u003c/h2\u003e \u003cdiv id=\"Sec7\" class=\"Section3\"\u003e \u003ch2\u003e2.4.1. Sentinel-2 Data\u003c/h2\u003e \u003cp\u003eWithin this study, the Sentinel-2 Level-2A (S2L2A) Collection 1 was employed. As part of the Copernicus program, the European Space Agency (ESA) has developed and launched the Sentinel-2 optical imaging mission, a constellation of polar orbiting satellites Sentinel-2A, B, and C, launched in June 2015, March 2017 and September 2024. Sentinel-2 provides near real-time open access data across multiple spectral bands at 10 m, 20 m and 60 m spatial resolutions. The revisit frequency of each single Sentinel-2 satellite is 10 days while the nominal two satellites configuration provides a revisit time of 5 days.\u003c/p\u003e \u003cp\u003eThe S2L2A collection consists in orthorectified Surface Reflectance (Bottom Of Atmosphere: BOA) images with Sen2Cor processor (Main-Knorn et al., 2017). Sen2Cor performs the atmospheric-, terrain- and cirrus correction of TOA (Top Of Atmosphere) L1C input data to generate BOA, optionally terrain- and cirrus corrected reflectance images; and additionally, Aerosol Optical Thickness-, Water Vapor-, Scene Classification Map and Quality Indicators for cloud and snow probabilities (Science Toolbox Exploitation Platform, 2024).\u003c/p\u003e \u003cp\u003eThe tiling system used by ESA for S2 is a Military Grid Reference System (MGRS), based on the Universal Transverse Mercator (UTM) projection. The MGRS system divides the Earth\u0026rsquo;s surface into 60 longitudinal zones. Each UTM zone is further divided into latitude bands of 8 degrees starting from the equator. Each 6\u0026deg; x 8\u0026deg; grid cell is subdivided into tiles of approximately 109.8 Km \u0026times; 109.8 Km. Adjacent tiles within the same UTM zone, overlap by around 4900 m while tiles along UTM zone borders exhibit even greater overlap with tiles from the neighboring zone (NASA, 2023). In this study, all images available covering Lebanon were considered from season 2017\u0026ndash;2018 (September 1st, 2017 to August 31st, 2018) to season 2024\u0026ndash;2025. Season 2016\u0026ndash;2017 was discarded from the analysis as only Sentinel-2a images were available during a large part of the season. Details about the number of images used in each season, as well as the tiles covering the study area, are shown in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Google Earth Engine (GEE) was used to access Sentinel-2 (S2) corrected images.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eNumber of images used in each season and the corresponding tiles covering the study area.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"8\" rowspan=\"9\"\u003e \u003cp\u003eS2L2A Collection Size\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSeason\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTotal Images\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eImages with Cloud Percentage ˂ 10%\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2017\u0026ndash;2018\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e542\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e263\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2018\u0026ndash;2019\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e659\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e277\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2019\u0026ndash;2020\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e662\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e342\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2020\u0026ndash;2021\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e663\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e427\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2021\u0026ndash;2022\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e676\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e371\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2022\u0026ndash;2023\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e650\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e298\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2023\u0026ndash;2024\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e658\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e316\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003e2024\u0026ndash;2025\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e835\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e423\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eS2 Tiles\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003e36SXB-36SXC-36SYB-36SYC-36SYD-37SBS-37SBT-37SBU\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section3\"\u003e \u003ch2\u003e2.4.2. Sentinel-2 Image Pre-processing\u003c/h2\u003e \u003cp\u003eA general workflow of this section is presented in \u003cb\u003eFigure S2\u003c/b\u003e (supplementary material). The Sentinel-2 images were stacked to be centered on the wheat crop season from first September to end August of each season. The region of interest (Lebanon) was clipped. Sentinel-2 Cloud Probability dataset was used to mask pixels with cloud probability higher than 65% and a linear temporal interpolation was applied to gap-fill the timeseries.\u003c/p\u003e \u003cp\u003eSnow and water mask were applied, using the Normalized Difference Snow Index (NDSI) and the Normalized Difference Water Index (NDWI) respectively as follows:\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:NDSI\\:\u0026gt;0.4\\:\\:\\:\\:\\:\\:;\\:\\:\\:\\:\\:\\:\\:\\:\\:NDSI=\\frac{RGB-SWIR1}{RGB+SWIR1}\\)\u003c/span\u003e \u003c/span\u003e \u003cem\u003e(Hall et al., 1995)\u003c/em\u003e\u003c/p\u003e \u003cp\u003e \u003cspan class=\"InlineEquation\"\u003e \u003cspan class=\"mathinline\"\u003e\\(\\:NDWI\\:\u0026gt;0.5\\:\\:\\:\\:;\\:\\:\\:\\:\\:\\:\\:\\:NDWI=\\:\\frac{Green-NIR}{Green+NIR}\\)\u003c/span\u003e \u003c/span\u003e \u003cem\u003e(McFeeters, 1996)\u003c/em\u003e\u003c/p\u003e \u003cp\u003eBlue, Green, Red, Red-Edge1, Near Infra-Red (NIR) and Short Wave Infra-Red (SWIR)1 and SWIR2 bands were selected along with the Normalized Difference Vegetation Index (NDVI) (Rouse et al., 1973) to create the input for the classification\u0026rsquo;s method. NDVI is a chlorophyll sensitive index, widely employed to quantify plant biomass and net primary productivity, to highlight vegetated content from other land cover types and to compare seasonal changes in vegetation growth (Myneni \u0026amp; Williams, 1994). Green, and Red-Edge bands correlate with chlorophyll and other leaf pigments (Delegido et al., 2011; Fern\u0026aacute;ndez-Manso et al., 2016), whereas SWIR is sensitive to water content and vegetation structure (Braga et al., 2021; Ceccato et al., 2002; Hunt \u0026amp; Rock, 1989). Previous studies also emphasized the importance of SWIR and Red-edge bands in classifying vegetation (Chaves et al., 2020). Sothe et al., (2017) identified these bands as decisive attributes to differentiate similar phenology. Furthermore, these bands were considered most informative for vegetation classification (Macintyre et al., 2020). Finally, the median values were computed on a monthly time scale to observe the temporal dynamics of the classes while avoiding GEE memory-limit issues. For each season, the dataset consists of 12 median monthly images, each with 8 spectral features (NDVI plus the 7 selected bands), resulting in a total of 96 features.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Image Classification and Assessment\u003c/h2\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003e2.5.1. Classification Algorithm\u003c/h2\u003e \u003cp\u003eThe general workflow of the image classification is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eConcerning the classification algorithm, a Random Forest (RF) was used. RF aggregates multiple decision trees (Breiman, 2001) and each decision tree casts a vote for a class label, and the forest outputs the class with the highest number of votes. This aggregation mechanism is referred to as majority voting (Probst \u0026amp; Boulesteix, 2018). Two parameters need to be defined when using RF: the number of features used at each node to generate a tree, denoted as \u0026lsquo;m\u0026rsquo; and the number of decision trees, denoted as \u0026lsquo;T\u0026rsquo;. In our study, \u0026lsquo;m\u0026rsquo; was set to the square root of the total number of features. Reducing \u0026lsquo;m\u0026rsquo; weakens individual trees but lowers their correlation, which in turn, strengthens the forest overall accuracy. RF uses bagging or bootstrap aggregating, meaning it creates training datasets by randomly resampling the original data with replacement, so that each tree sees a different version of the data. However, this means that some features may appear multiple times, while others might be underrepresented and not appear at all (Rodriguez-Galiano et al., 2012). A study showed that the biggest performance gain in RF happened within the first 100 trees, however, it is important to note that other parameters can also influence the results, such as the sample size, the number of variables (Probst \u0026amp; Boulesteix, 2018). Breiman, (1996) showed that increasing \u0026lsquo;T\u0026rsquo; leads to the convergence of the generalization error without overfitting, thanks to the strong law of large numbers (Feller, 1991). To ensure good coverage of all observations without excessive computational cost, a forest of 300 decision trees was constructed to classify the input image. This number was determined through a sensitivity analysis, in which 2019\u0026ndash;2020 data were used to train a classifier with an increasing number of trees. The trained classifier was then applied to 2018\u0026ndash;2019 season, and its performance was evaluated. The results are shown in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Although the results are quite stable independently of the number of trees, the highest overall accuracy and kappa coefficient values were recorded for T\u0026thinsp;=\u0026thinsp;300, therefore, it was selected as final choice.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOverall Accuracy and Kappa coefficient of the Random Forest classifier with increasing number of trees.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNumber of Trees\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e30\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e200\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e300\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e500\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eOA\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eKappa\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo assess the temporal robustness of our classification strategy, we performed crop type mapping over multiple wheat growing seasons from 2017\u0026ndash;2018 to 2024\u0026ndash;2025. Rather than training a new classifier for each year, we aimed to train a single RF model using one season data, which would then be applied to all other seasons. This approach is intended to mimic the production workflow for an operational monitoring service, where in situ data collection and model retraining cannot be carried out every year, particularly in situations where collecting reference points is difficult due to crises, financial or logistical constraints. By comparing the accuracy obtained with this fixed classifier to that of annually retrained models, we evaluate the expected degradation in performance over time and quantify the trade-off between operational efficiency and classification accuracy. The classifiers were trained separately on each dataset (2017\u0026ndash;2018, 2018\u0026ndash;2019, 2019\u0026ndash;2020, and 2022\u0026ndash;2023), and their accuracies were assessed on the remaining three seasons for which we had reference datasets. The 2019\u0026ndash;2020 dataset was divided into training (70%) and testing (30%) subsets. After performing the accuracy assessment, the trained classifier was exported and used to classify the multi-temporal input images from 2017\u0026ndash;2018 to 2024\u0026ndash;2025. Eight classification maps for Lebanon covering the seasons from 2017\u0026ndash;2018 until 2024\u0026ndash;2025, at a spatial resolution of 10 m were generated.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e \u003ch2\u003e2.5.2. Classification Assessment and Consolidation\u003c/h2\u003e \u003cp\u003eThe map assessment is performed using classical metrics (confusion matrix, overall, producer and user accuracy, kappa coefficient) as detailed in the supplementary material Section S3. As an additional assessment step, the available wheat fields from the four seasons were used to evaluate how accurately the predictions made by the classifier align with the true data. The wheat parcels for each season were overlaid on the corresponding classification maps, and using the Zonal Statistics tool in QGIS, the predominant mapped class within each parcel polygon was recorded. The number of correctly mapped parcels was then calculated to determine the percentage of agreement between the classifier output and the ground truth.\u003c/p\u003e \u003cp\u003eSeveral post-classification assessment methods were implemented. A general workflow of this part is presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFirst, change maps and transition matrices were established to evaluate class stability, consistency, and the continuity of the maps. These matrices help understand class behavior, detect trends, and identify real-world changes. Post-Classification comparison is a method of change detection, which requires the comparison of independently produced classified images (Gordon, 1980; Howarth \u0026amp; Wickware, 1981; SINGH, 1989). To this objective, the pixels that remained in the same land cover class, as well as those that transitioned to a different class in the following season were identified. Transition probabilities were calculated for each class based on the class total number of pixels in each season. By identifying unlikely transitions, we assessed the classification errors behind these unreasonable changes. The two main \u0026ldquo;unlikely\u0026rdquo; transition identified is the change from forest or built-up to another class. Within this context, we used repetition across years to reinforce confident forest and built-up classifications, correcting the transitions that seemed unreasonable within these two classes. For each pixel across the 8 yearly maps, we checked how many times it was classified as built-up or forest. If a pixel was classified as built-up 4 or more times, we corrected it to built-up in all years where it had a different class. Similarly, if a pixel was classified as forest 4 or more times, it was corrected to forest in the years where it had another class. Indeed, even if new buildings are constructed or new tree plots can be planted, this will affect only a relatively small area. If a pixel was classified as built-up in one year but appeared less than 3 times across the 8 maps, it was labeled as uncertain built-up. If a pixel was classified as forest in one year but appeared less than 3 times, it was labeled as uncertain forest.\u003c/p\u003e \u003cp\u003eThe RF classifier provides the estimated probabilities for each land cover class for every pixel in each season. Within this study, we also computed and used the confidence, defined as the probability of the predicted class, and the score margin, defined as the difference between the top two class probabilities, in order to select only confident predictions, while retaining flexibility by controlling how strongly the wheat class needs to dominate over the second-highest class (based on score margin). Pixels with a confidence higher than 50% and a score margin higher than 0.3 were extracted and wheat area was calculated accordingly. It is important to note that the classifier\u0026rsquo;s confidence score represents its internal certainty in each prediction based on patterns learned from the training data. However, it does not indicate real-word accuracy, which can only be determined through independent validation.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e2.5.3. Sample-based Area Estimation and Accuracy Assessment\u003c/h2\u003e \u003cp\u003eAutomated classifications, especially in case of large areas, will inevitably contain some errors and biases. Errors can arise from different factors such as noise, unclear class boundaries, gaps in the time series. Biases can result from the methods used for classification and change detection, as well as from human choices when collecting training data, which are often based on available reference data. To derive unbiased area estimates and quantify classification uncertainty, we adopted the Random Stratified Sample (RSS) approach proposed by Olofsson et al., (2013, 2014). This method replaces the pixel-counting procedure, which assumes the map is error-free, by a statistically rigorous estimation of class proportions derived from a reference sample.\u003c/p\u003e \u003cp\u003eGiven that reference data does not differentiate between confident and non-confident wheat classification. The Random Stratified Sample (RSS) method was applied on the raw classification map (before selecting the confident wheat pixels) in order to estimate the errors on the estimated area of wheat crops. Then, the confident wheat area and the uncertainties were extracted from the total wheat estimated area. Overall, this method ensures that reported land-cover areas and their uncertainties are statistically unbiased, and comparable across studies, representing current best practice for area estimation in land-cover mapping. The computing details are provided in the supplementary material Section S4.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Classification Maps Assessment\u003c/h2\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.1.1. Statistical Metrics\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e reports the classification metrics for the four seasons with respect to in-situ measurements, and for the cross-season runs where a classifier trained on one season is applied to the other seasons. The classifiers show on average high performance with Overall Accuracy (OA) and Cohen\u0026rsquo;s Kappa Coefficient (Kappa) values mostly above 0.80. The model trained on 2019\u0026ndash;2020 exhibits the highest temporal transferability, with overall accuracy values ranging between 0.90 and 0.94 and Kappa coefficients between 0.88 and 0.92 when applied on adjacent seasons 2017\u0026ndash;2018 and 2019\u0026ndash;2020. The good correspondence between OA and Kappa also suggests a strong agreement between predictions and reference labels beyond what would be expected by random identification. In fact, Kappa, which corrects OA for random agreement, remains close to OA values (Congalton \u0026amp; Green, 2019; Foody, 2002; Landis \u0026amp; Koch, 1977). By contrast, the classifier trained on 2022\u0026ndash;2023 performs less consistently when transferred backward in time (OA\u0026thinsp;\u0026le;\u0026thinsp;0.80; Kappa\u0026thinsp;\u0026le;\u0026thinsp;0.75). Overall, the observed decrease in both OA and Kappa highlights the change of land-cover spectral characteristics and probably of land management practices as well potentially linked to an anomalously dry climate in the Eastern Mediterranean region at this period (cf. the precipitation amount for season 2022\u0026ndash;2023 in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e below).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAccuracy assessment metrics for the four classifiers tested.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTrain/Test\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2017\u0026ndash;2018\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2018\u0026ndash;2019\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2019\u0026ndash;2020\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2022\u0026ndash;2023\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eClassifier 2017\u0026ndash;2018\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.96\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.86\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.87\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.79\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eClassifier 2018\u0026ndash;2019\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.91\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.95\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.90\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.79\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.74\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eClassifier 2019\u0026ndash;2020\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.92\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.90\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.94\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.92\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.80\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.75\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eClassifier 2022\u0026ndash;2023\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.87\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.79\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.76\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003eOA\u0026thinsp;=\u0026thinsp;0.86\u003c/p\u003e \u003cp\u003eKappa\u0026thinsp;=\u0026thinsp;0.83\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e displays the Producer Accuracy (PA) and User Accuracy (UA) for each class using the 2019\u0026ndash;2020 that exhibited the best overall performance based on global statistical metrics (OA and Kappa). Overall, class accuracies show strong performance for most land-cover types across years. Wheat and built-up areas consistently exhibit high PA and UA (\u0026gt;\u0026thinsp;0.90 in most years). Forest also exhibits high UA (\u0026gt;\u0026thinsp;0.90), although PA fluctuates slightly. Bare soil shows more variability across years, likely reflecting differences in soil moisture, tillage state and vegetation residues influencing its spectral separability. The most pronounced decline occurs for the vegetation class in 2022\u0026ndash;2023 as PA drops sharply to 0.41 and UA to 0.70, suggesting a substantial change in the spectral response or separability of this class as already observed on the global metrics above. This was also partly expected as the vegetation class gather all type of vegetation that is not classified as forest or wheat meaning the intra-class spectral variability is high by definition.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eProducer Accuracy (PA) and User Accuracy (UA) for each cover class in the four seasons with reference data using the 2019\u0026ndash;2020 classifier.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWheat Season\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eWheat\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eVegetation\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eForest\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eBare soil\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eBuilt-up\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e2017\u0026ndash;2018\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.91\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e2018\u0026ndash;2019\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.92\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.95\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.87\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e2019\u0026ndash;2020\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.97\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003e\u003cb\u003e2022\u0026ndash;2023\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.96\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eUA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.91\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe analysis of compliance between the classifier and the ground truth data, using all available wheat fields (from CNRS-L and DGCB), showed that on average, 81% of the fields were correctly mapped by the classifier across the four seasons (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). The remaining fields were generally misclassified as vegetation. This result is somewhat expected, given that UA of the vegetation class was lower than that of wheat, indicating that a larger proportion of the areas classified as vegetation did not actually belong to this class.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eCompliance % between the surveyed wheat fields and the classification maps.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSeason\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2017\u0026ndash;2018\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2018\u0026ndash;2019\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2019\u0026ndash;2020\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2022\u0026ndash;2023\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eCompliance %\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAs a conclusion and hereafter, the classifier 2019\u0026ndash;2020 is chosen as the best candidate for classification on the whole time series. Considering wheat that is the subject of the present study, accuracies remains very high for most the years with a slight decline in 2022\u0026ndash;2023. Overall, these results indicate that the model trained on 2019\u0026ndash;2020 data generalizes quite well and will be chosen hereafter as the classifiers for mapping land cover type along the whole time period. Based on this classifier, seven wheat maps for Lebanon covering the seasons from 2017\u0026ndash;2018 until 2023\u0026ndash;2024, at a spatial resolution of 10 m were generated.\u003c/p\u003e \u003cp\u003eFinally, as a complementary way to assess classification accuracy, \u003cb\u003eFigure S5\u003c/b\u003e (supplementary material) displays the mean confidence and the mean score margin for each class across the years based on the 2019\u0026ndash;2020 classifier. Values range between 45 and 70% depending on the class and the seasons. However, such raw values must be interpreted with caution: they are not true posterior probabilities unless the model is well calibrated (Berio Fortini et al., 2024). By contrast, the score margin (the difference between the highest vote proportion for the predicted class and the second-highest vote proportion) offers a more robust indicator of separation between classes. Empirical studies in remote sensing classification demonstrate that enforcing a minimum score margin around 0.30\u0026ndash;0.40 markedly improves class-label reliability (Aitkenhead \u0026amp; Dyer, 2007). Although the lowest performance is observed for the vegetation class, our results show margins consistently exceed 0.30 for all classes proving clear separability between the top two classes\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e3.1.2. Transition Matrices\u003c/h2\u003e \u003cp\u003eThe analysis of classes transitions over consecutive seasons reveals an acceptable degree of stability, with some changes in class distribution (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). For wheat, most observed transitions involve shifts to vegetation class probably because of crop rotation. Some transitions seemed slightly unreasonable, particularly the wheat-built-up and forest-vegetation ones; therefore, further investigation was conducted to evaluate the accuracy of these probabilities. An average of 10% transition from wheat to built-up was depicted for instance, which is unusually high. This is likely due to a classification error. Similarly, forest class stood out with some anomalies, showing unexpected transitions (24% of forest transitioned to vegetation). This can highlight real forest degradation or classifier confusion. As mentioned before, the vegetation class is too broad (high internal variance). Moreover, forests in Lebanon are significantly fragmented with only 10% being over 200 ha. Natural forest associations consist of pure and mixed stands including different species, notably Cedars of Lebanon, various types of Junipers (Talhouk et al., 2001). This can create overlap between classes as observed by the confidence level and score margin from \u003cb\u003eFigure S5 (supp. material)\u003c/b\u003e, where vegetation class exhibits the lowest values. Therefore, the 24% transition from forest to vegetation is likely inflated by this ambiguity, and could be partly artefact, not entirely pure ecological change. Within this context, the classification maps were corrected based on this unlikely transition.\u003c/p\u003e \u003cp\u003e\u003cstrong\u003eTable 7\u003c/strong\u003e reports the change before and after correcting the built-up and forest pixels in terms of unlikely transitions (%). To this objective, we analyzed the repetition of built-up and forest pixels across the seven seasons and we set a threshold of at least 4 repetitions over 8 seasons for pixels to be considered true built-up/forest pixels. After correcting the new misclassified pixels, the transition probabilities were significantly improved with drastic drops of unlikely transition: forest-vegetation going from 26% during the 2018 to 2019 transition to 3%. Likewise, the wheat-built-up was 20% from 2021 to 2022 and dropped to 5%.\u003c/p\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab7\" border=\"1\"\u003e\u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv\u003e\n \u003ctable id=\"Tab8\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 7\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eTransition probabilities (%) before and after correction of forest and built-up pixels. \u003cem\u003e(2018→2019 means the transition from season 2017–2018 to 2018–2019).\u003c/em\u003e\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"8\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTransition\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2018\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2019\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2020\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2021\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2022\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e2024\u003c/p\u003e\n \u003cp\u003e\u003cem\u003e→\u003c/em\u003e\u003c/p\u003e\n \u003cp\u003e2025\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eWheat → Built-up\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 → 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e20 → 5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 → 5\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eVegetation → Forest\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 → 4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 → 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e9 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e7 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eForest → Vegetation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26 → 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 → 8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21 → 9\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24 → 7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32 → 6\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e28 → 10\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35 → 8\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eBuilt-up → Wheat\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 → 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11 → 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4 → 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2 → 0\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\"\u003e\n \u003ch2\u003e3.2. Wheat Maps and Wheat Area Estimation\u003c/h2\u003e\n \u003cp\u003eFor visualization purposes, Fig. 8 shows wheat pixels in a small agricultural region located in the Bekaa, between Zahle and the West Bekaa, the breadbasket of Lebanon, as an illustration of the classification maps. The complete maps are provided in \u003cstrong\u003eFigure S6\u003c/strong\u003e (supplementary material). The wheat/non-wheat masks exhibit a good spatial consistency: field boundaries are well resolved and parcel geometries can be clearly observed. In addition, the interannual variability proved to be high with only a small fraction of parcels remains under wheat in two consecutive seasons. This is also consistent with local crop-rotation practices. Indeed, multiple sources explain that Bekaa farmers frequently rotate wheat with potatoes and other vegetables (Caiserman et al., 2019; Jaafar et al., 2024; Tawk et al., 2019).\u003c/p\u003e\n \u003cp\u003eWheat area fractions with 95% confidence intervals were estimated using a random stratified sample for four seasons where reference points were available. A linear trendline (R² = 0.9) was fitted to examine the relationship between the wheat fraction estimated from pixel counts and that estimated from the RSS. The trendline is shown in \u003cstrong\u003eFigure S7\u003c/strong\u003e (supplementary material). The trendline equation was then used to estimate the fraction in seasons where samples were not available. For the confidence intervals, the average error rate (14%) was applied to determine error intervals for the remaining seasons. The confident wheat fraction, with 95% confidence intervals, was then derived from the estimated values based on the fraction of confident pixels in each season. Finally, the area was calculated in hectares. The comparison between the two estimation methods results (pixel counts and RSS) showed that the mapped wheat area was underestimated in all seasons except for 2018–2019 (Table 8) with a negative bias varying from − 4,272 ha (almost 14% of the wheat cropped area) to -657 ha (1%) in 2023–2024.\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab9\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 8\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eWheat area based on the random stratified sample and pixel count. In grey, are reported the seasons when the linear trendline was used by lack on in situ data (see text).\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"5\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSeason\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePixel Count Area (ha)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eRandom Stratified Sample Area (ha)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eBias (ha)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eBias (%)\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2017–2018\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25,536\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29,835\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-4,272\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e14\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2018–2019\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29,469\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27,212\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e+ 2,257\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2019–2020\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29,039\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30,432\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-1,393\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2020–2021\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e53,005\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e53,751\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-746\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2021–2022\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32,604\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35,449\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-2,845\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e8\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2022–2023\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e66,383\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e69,194\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-2,811\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2023–2024\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51,142\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51,799\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-657\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e2024–2025\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35,491\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e37,642\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e-2,151\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e6\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\"\u003e\n \u003ch2\u003e3.3. Interannual Variability of Wheat Area\u003c/h2\u003e\n \u003cp\u003eWe observed that the harvested area fluctuated over the seasons (Fig. 9). Wheat area decreased by 10% from 2017–2018 to 2018–2019, then increased again at a similar rate in 2019–2020. It rose sharply by 43% in the following season (2020–2021). In 2021–2022, wheat area dropped significantly by about 52%, then increased again in 2022–2023, reaching the highest level of the study period. However, it declined once more by around 34% in 2023–2024 and continued decreasing by 38% in 2024–2025. Overall, the wheat area showed alternating rises and falls, ranging from a minimum of 27,212 ha in 2018–2019 to a maximum of 69,194 ha reached in 2022–2023, reflecting a pronounced interannual variability (coefficient of variation: 35%).\u003c/p\u003e\n \u003cp\u003eThe analysis of the average precipitation in the main wheat producing regions during wheat season showed no correlation (-0.01) between wheat area and precipitation (Fig. 9). Particularly, in the seasons where we observed sharp variations in wheat area (2021–2022, 2022–2023 and 2023–2024), precipitation did not vary in the same pattern or intensity. Furthermore, the 2020–2021 wheat season began with a pronounced drought, as SPEI-90d values fell sharply in September 2020 (provided in the supplementary material \u003cstrong\u003eFigure S8\u003c/strong\u003e). This severe moisture deficit persisted for several months, yet wheat cultivation area during this cycle was higher than in the previous season, despite the fact that hygrometric conditions had been more favorable in 2019–2020. Seasonal precipitation is a key driver of wheat growth, particularly early in the season when adequate rainfall replenishes soil water reserves and supports crop establishment; however, its influence in Lebanon is partly moderated because many farmers buffer irregular rainfall through supplemental irrigation. Over 80% of wheat plots are supplementary irrigated (Nasrallah et al., 2019), unlike other countries, such as Morocco, where approximately 83% of wheat is grown under rainfed conditions (Lago-Olveira et al., 2024). Farmers there, typically wait to assess rainfall levels before sowing, and may abandon planting if rainfall appears insufficient, which is not the case in Lebanon. According to a past survey, 62% of West Bekaa farmers irrigate once or twice, while the remaining farmers irrigate three or four times (Tawk et al., 2019). However, it’s important to note that although precipitation itself does not seem to influence wheat area, water availability can still indirectly shape farmers’ decisions. In addition, during periods of economic or political crisis, the feasibility of supplemental irrigation becomes more constrained: high fuel prices raise the cost of pumping groundwater, the price of purchased irrigation water may increase, and labor requirements for manually irrigated fields can become prohibitive. Under these conditions, wheat water requirements may not be fully met, consequently affecting yields (Achli et al., 2025), and potentially influencing whether farmers choose to continue cultivating wheat or shift to other crops. The macro-economic factors beyond climate that may influence harvested area are discussed in \u003cstrong\u003esection 4.2\u003c/strong\u003e.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e \u003ch2\u003e4.1. Wheat Mapping Approach\u003c/h2\u003e \u003cp\u003eAs already observed on the wheat mask on the Bekaa Valley (Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e), the transition matrices showed that, on average, 22% of wheat transitioned to vegetation, from one season to another, which can be partly explained by crop rotation practices in Lebanon. Unlike monoculture, diversified rotations can increase the yield. Moreover, including legumes stimulates soil microbial activities, adds nitrogen to the soil, increases soil organic carbon stocks, and enhances soil health, thereby improving fertility for the next crop. Rotations also can help improve soil structure and prevent issues like compaction (Volsi et al., 2022; Yang et al., 2024). Other than rotation, the transition to vegetation or bare soil can also result from crises factors that may influence a farmer\u0026rsquo;s choice to cultivate or not. This was further supported by the analysis of transitions between bare soil/vegetation and wheat. Table\u0026nbsp;\u003cspan refid=\"Tab10\" class=\"InternalRef\"\u003e9\u003c/span\u003e reports the wheat gain and losses during the transition between two consecutive years during the study period. Two critical seasons stood out: 2021\u0026ndash;2022 and 2023\u0026ndash;2024, during which we observed a sharp increase in wheat loss and a significant decrease in wheat gain, deviating from the patterns observed in the previous seasons. Conversely, in 2022\u0026ndash;2023, wheat loss declined to its lowest level, while wheat gain increased. These results indicate that new factors emerged and influenced the farmers\u0026rsquo; crop choices. Specifically, during 2021\u0026ndash;2022, 2023\u0026ndash;2024 and 2024\u0026ndash;2025, farmers faced severe financial and political constraints, respectively. In contrast, the relative stabilization of the Lebanese economy in 2022\u0026ndash;2023 encouraged farmers to reinvest in wheat production. These variations are further discussed in section \u003cb\u003e4.2 below\u003c/b\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab10\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 9\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eWheat gain and loss probabilities over the study period. (2018\u0026rarr;2019 means the transition from season 2017\u0026ndash;2018 to 2018\u0026ndash;2019).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSeasons Transitions\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2018\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2019\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2019\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2020\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2020\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2021\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2021\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2022\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2022\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2023\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e2023\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2024\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e2024\u003c/p\u003e \u003cp\u003e\u0026rarr;\u003c/p\u003e \u003cp\u003e2025\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWheat \u0026rarr;\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003eVegetation (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eWheat \u0026rarr;\u003c/b\u003e\u003c/p\u003e \u003cp\u003e\u003cb\u003eBare soil (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eVegetation \u0026rarr; Wheat (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eBare soil \u0026rarr; Wheat (%)\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eExtracting all wheat pixels identified by the classifier can include low-confidence predictions, which may lead to an overestimation of the wheat area. Conversely, selecting only pixels with a confidence level above 80% risks underestimating the actual wheat area. Moreover, as confidence and score margin levels vary, the expected wheat pixels also change (Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e). Therefore, achieving a reliable estimate requires selecting pixels based on a combination of confidence and score margin to ensure that only reasonably confident pixels are included. By this, we mean pixels that are not selected randomly and are well differentiated from other classes. For this reason, we selected pixels with a confidence greater than 50% and a score margin above 0.3. Finally, interestingly enough, the interannual dynamics remain similar regardless of the chosen confidence level or score margin threshold, meaning, that the choice of the thresholds is not critical for this study.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eArea estimation based on simple pixel-counting is biased because it does not take classification errors into account; as a result, the area estimates are erroneous. Misclassified pixels lead to misestimation of class areas (Olofsson et al., 2014). Indeed, when comparing wheat area derived from pixel counts to the area estimated using the RSS approach, our results showed a systematic bias. On average, an underestimation of the wheat area was observed across all seasons. This finding reinforces the importance of using RSS-based area estimation to obtain more reliable results.\u003c/p\u003e \u003cp\u003eThe presence of errors affects the width of confidence intervals for the area estimates, the larger the errors, the greater the uncertainty. Particularly, omission error in maps can add substantial uncertainty to area estimates (Olofsson et al., 2020). In the 2018\u0026ndash;2019 season, PA which is the complement of the omission error (Gallego et al., 2010) was the highest, which means the omission error in the wheat class was the lowest. This can explain the narrow confidence interval observed for this season (9%). For the remaining three seasons, the confidence intervals were wider. Based on the OA and PA, we expected the confidence interval of 2022\u0026ndash;2023, to be the highest and indeed, that was the case (18%). However, our results showed that the error rate in 2019\u0026ndash;2020 (17%) was higher than in 2017\u0026ndash;2018 season (14%), despite the higher PA in 2019\u0026ndash;2020. This difference can be explained by the sample size. For the 2019\u0026ndash;2020 season, the sample represented 30% of the sample used for 2017\u0026ndash;2018, which can explain the higher confidence interval for this season.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section2\"\u003e \u003ch2\u003e4.2. Link between Multiple Crises and Wheat Growing Areas\u003c/h2\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e displays the wheat areas over the whole country and associated uncertainties together with the timing of the main crisis. The pre-crisis period (2017\u0026ndash;2018 and 2018\u0026ndash;2019) was overall stable with wheat areas remaining around 28,000\u0026ndash;30,000 ha. The predictable economic conditions allowed farmers to plan cultivation decisions based on established cost-revenue relationships and access inputs through functional markets. The relatively low wheat areas during this period, likely reflect structural constraints in Lebanon\u0026rsquo;s wheat sector, where wheat often remains a rotation crop competing with higher-value alternatives such as potatoes or vegetables (Tohme Tawk et al., 2019). For instance, Nasrallah et al., (2019) stated that the relatively low wheat areas during season 2016\u0026ndash;2017 (outside of our study period) was related to the choice of farmers to plant potatoes during this season. In addition, FAO GIEWS, (2018) reported unfavorable early-season conditions during winter season 2017\u0026ndash;2018, which could have further constrained sowing decisions and reinforced low baseline areas prior to the 2019 economic crisis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWheat area slightly increased by 11% in 2019\u0026ndash;2020, notably after the beginning of the financial crisis in Lebanon (Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e). During this crop season, the Lebanese Agricultural Research Institute (LARI) distributed wheat seeds free of charge upon the Ministry of Agriculture (MoA) request. Wholesalers also provided farmers with financial resources early in the season to guarantee cash flow that the farmers would repay in-kind (as a portion of the output) at the time of the harvest (FAO, 2020). Moreover, local wheat production was still supported by the government. All these factors were still encouraging farmers to keep cultivating wheat.\u003c/p\u003e \u003cp\u003eThe following boom-bust cycle between 2020 and 2022 reveals the combined influence of strategic priorities and economic constraints on farmer decision-making under extreme crisis pressure. In 2020\u0026ndash;2021, the expansion of wheat area by 83% with regards to the three previous year average, occurred despite three concomitant crises: the deepening of the financial collapse, the lock down due to the pandemic and the Beirut port explosion. Although, the production cost increased by 40% during this season (Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e12\u003c/span\u003e), the area increased, violating conventional agricultural economics where area is expected to decline in response to high costs, and highlighting that food security crisis can trigger powerful counter-cyclical responses. In fact, the destruction of Lebanon\u0026rsquo;s major grain silos in Beirut port in August 2020 created immediate threat of bread shortages (Breisinger et al., 2023), prompted coordinated government and international donor emergency interventions (FAO, USAID, World Bank). The global markets were disrupted by the COVID-19 lock down and the prices were rising in Lebanese market (food inflation rose from 11% to 205%). Food security concerns overrode economic calculations, triggering return migration from urban areas to family farms, where wheat cultivation provided food self-sufficiency, supplemental cash income, and some independence from the failed financial system. Furthermore, a new local domestic pasta manufacturer has started operations, using 100% Lebanese durum wheat. This increased the opportunity for wheat farmers to sell their production, further explaining the increase in wheat area.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn 2021\u0026ndash;2022, growing wheat became a major challenge, as the production cost became fundamentally untenable, with LBP costs exploding by 327% (the steepest single-year increase in the entire study period). Since 2021, fuel prices have been increasing, rising from 315,000 Lebanese Pounds for 20L of Diesel in 2021 to 841,000 Lebanese Pounds in 2022 (IPT Group, 2025). Furthermore, fertilizers price increased drastically. According to Lebanese Customs, the price of one ton of fertilizer, which had hovered around 253 USD from 2017 to 2020, sharply increased to 385 USD in 2021, and reached 596 USD in 2022, notably after the conflict between Russia and Ukraine began in February 2022 (Lebanese Customs, 2024) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). According to data provided by Maison d\u0026rsquo;Agriculteur, the average price of local wheat seeds has increased by approximately 22% in 2021. In contrast, the average price of imported wheat seeds has remained stable in this period, at around 1,100 USD per ton, largely constrained by the limited purchasing capacity of farmers. The 252% deterioration in LBP-USD exchange rate meant that every USD-priced input became approximately 3.5 times more expensive in LBP terms, creating a fundamental mismatch between farmer revenues and input costs which were effectively denominated in USD given import dependency. At the same time, the real burden, measured in bread bundle equivalent, expanded by 129%, while the food price inflation peaked at a hyperinflationary 440% (the highest level recorded in Lebanese history). As a result, farmers needed 2.3 times more purchasing power than the previous season to simply cover production costs, representing an impossible demand. Furthermore, the crisis was directly linked to the government\u0026rsquo;s inability to sustain subsidies, confirmed by the use of foreign reserves to subsidize vital commodities like wheat (Abou Ltaif et al., 2024). Consequently, for this season, the government was no longer able to support the farmers, causing them to lose an important marketing channel, and forcing them to find alternative market routes, capable of absorbing their high production costs, to allow even minimal profits. Due to the financial collapse, farmers also lacked access to credits to purchase new inputs and were requested to pay old arrears, and to purchase inputs in cash either at face value in US dollars or in Lebanese pounds using unofficial exchange rate, translating into an increase in agricultural input costs compared to the previous seasons, based solely on exchange rate fluctuation (FAO, 2020). This situation affects particularly, small farmers who lack sufficient capital and cannot sustain operations without immediate cash flow. Changing their agricultural practices also carries the risk of lowering their productivity, as higher yields were mainly driven by the application of improved varieties, irrigation, fertilizers, and pesticides (Tadesse et al., 2017; Tita et al., 2025). All these elements made wheat cultivation a losing scenario, which explains the sharp decrease in harvested area (52%).\u003c/p\u003e \u003cp\u003eA remarkable recovery happened in 2022\u0026ndash;2023, despite continuously increasing production costs. Interestingly, the observed increase in LBP-denominated costs was driven mainly by exchange rate deterioration rather than by international price inflation. Indeed, a decline in USD costs (provided in the supplementary material \u003cb\u003eFigure S9\u003c/b\u003e) was noted. Furthermore, the February 2022 onset of the Russia-Ukraine war, disrupted global wheat markets and caused a spike in wheat prices (Kuhla et al., 2024). High international wheat prices made domestic wheat production economically attractive and when combined with exchange rate depreciation, amplified nominal profitability in LBP terms. Food inflation also moderated to 310.8% (still extreme but decelerating), partially restoring purchasing power and the real burden (bread bundle equivalent) declined. Indeed, in 2023, after four years of severe financial crisis, a small degree of stabilization began. Lebanese economy began partially switching to dollarization, maintaining a minimum level of stability in prices (Banque du Liban, 2023). Farmers likely started to adapt to this post-crisis environment, and optimize input choices, helping to explain the expansion of wheat area in this season.\u003c/p\u003e \u003cp\u003eUnfortunately, as the wheat area began to increase, the offensive Israeli war on Lebanon started in October 2023. This renewed instability explains the decrease in wheat area, particularly in the south, which was the most affected region, recording the highest number of attacks (provided in the supplementary material \u003cb\u003eFigure S10\u003c/b\u003e), and witnessed the highest average loss percentage of 47% and 61% in 2023\u0026ndash;2024 and 2024\u0026ndash;2025 seasons, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e13\u003c/span\u003e). By December 2023, caretaker Minister of Public Health Firas Abiad announced that around 900,000 people were displaced (OCHA, 2024). Farmers may have abandoned their land due to displacement, or they risked to be injured or dead from shelling, airstrikes, while attempting to plant or harvest, and even though the production cost was more favorable than the previous season, farmers simply could not cultivate land in active conflict zones regardless of input costs, output prices, or profitability considerations. This is supported by the fact that the transition from wheat to bare soil increased from 9% to 24% in 2024\u0026ndash;2025 season (see \u003cb\u003eTable\u0026nbsp;10\u003c/b\u003e above).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eOverall, based on this general analysis, the choice to grow wheat appears to result from multiple influencing factors. In time of crisis, understanding this choice becomes more complex due to the variability of the factors between regions and the diverse ways farmers think and respond. Previous observations support this point of view, suggesting that individual farmers\u0026rsquo; decisions and responses to the changing agricultural environment, such as biophysical conditions, socio-economic factors, and policy interventions, collectively shape agricultural patterns over time. Therefore, the choice to grow a certain crop is not a simple decision but a multifaceted process, especially under risk. In Vietnam, for example, risk aversion is likely linked to staying in rice monoculture, despite the strong beliefs and desires to shift to production of other crops because of rice low profitability. In contrary, knowledge of new alternatives increase the likelihood that farmers will switch from rice monocultures to other crops (Le et al., 2024). In Ghana, 41% of farmers preferred to grow cocoa instead of food crops, because of high market risk for food crops, as the cocoa sector offers a ready market and price hikes. However, others avoid cocoa farming due to perceptions of lower potential returns compared to food crops. For them, cocoa requires higher input and labor costs, coupled with the longer maturity period, making food crop farming more profitable (Hashmiu et al., 2022). In Nigeria, violent conflicts are major factors shaping agriculture. They often reduce the amount of land harvested, largely because farmers fear imminent attacks. In addition, cropping patterns are affected, with a substantial shift from long-term crops, such as perennials, to short-term crops (Amare et al., 2025). This is the case also in Ukraine where a drop of 25% of wheat productivity was observed in the areas invaded by Russia (Antonenko et al., 2024).\u003c/p\u003e \u003cp\u003eIn the case of Lebanon, previous studies have also highlighted that the choice to grow wheat isn\u0026rsquo;t constant, but rather but rather depends on the environment in which farmers operate. Socio-economic surveys conducted in 2017, showed that major farmers in the Bekaa region chose wheat because of its profitability. Since the government purchased the entire production each year, wheat was considered a secure crop compared to others, such as vegetables, which had fluctuating and unreliable prices from one season to another (Caiserman et al., 2019). Furthermore, Nasrallah et al. (2018), attributed the decrease in wheat area from 2016 to 2017 to various factors, including agricultural practices, corrupt subsidy policies, the Syrian war, and marketing policies.\u003c/p\u003e \u003cp\u003eSeveral reports have investigated the state of agriculture in Lebanon and the adaptation techniques adopted by farmers, such as decreasing expenses for agricultural inputs (United Nations, 2023) by choosing substitution for cheaper alternatives (replacing fertilizers by locally-sourced substitutes, such as manure or compost, or a composite of local industrial phosphate mixed with local organic fertilizers) or decreasing quantities of inputs used, cultivating smaller areas (FAO, 2020). However, this overview remains general and somewhat limited for several reasons. First, the intensity of the crisis varies significantly across regions, leading to different local dynamics. Second, agricultural environments differ among farmers, influencing their capacity to adapt and make production choices. Most importantly, crop yield has not been estimated, even though it is a key factor since productivity directly affects the profit farmers can obtain. Therefore, while this analysis offers valuable insights, it remains a preliminary step.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Conclusion and Perspectives","content":"\u003cp\u003eOur study successfully mapped winter wheat areas across Lebanon over successive seasons, from the 2017\u0026ndash;2018 season up to 2024\u0026ndash;2025, using remote sensing technology, and analyzed the inter-annual variability of these areas in the context of the country's multiple, overlapping crises. The outcomes of this research are both technical and agricultural.\u003c/p\u003e \u003cp\u003eFrom a technical perspective, the RF classifier, used in our study, demonstrated strong performance and temporal transferability across the time series. The model trained on 2019\u0026ndash;2020 data exhibited the highest consistency and was chosen for classification across the whole time period (2017\u0026ndash;2018 to 2024\u0026ndash;2025). Overall Accuracy for the estimation reached 87%. Wheat classification specifically showed consistently high Producer Accuracy (PA) and User Accuracy (UA) for most years, although a slight decline was observed in 2022\u0026ndash;2023. The results confirm that classification maps alone can be biased if relying solely on pixel counts for area estimation. The statistically rigorous Random Stratified Sample (RSS) approach was applied to derive unbiased estimates and quantify classification uncertainty. The comparison between methods showed that the mapped wheat area was underestimated by pixel counts in all seasons except 2018\u0026ndash;2019, with the bias reaching up to 4,272 ha. Therefore, incorporating reference samples is necessary for more accurate area estimation.\u003c/p\u003e \u003cp\u003eFrom agricultural and socio-economical perspective, our study successfully monitored the fluctuation of wheat area, indicating that farmers\u0026rsquo; decisions regarding wheat cultivation were significantly influenced by multiple factors related to the multi-factorial crisis threatening Lebanon. Wheat area generally increased until 2020\u0026ndash;2021, but a sharp decrease of about 52% occurred in the 2021\u0026ndash;2022 season, followed by a significant recovery in 2022\u0026ndash;2023, reaching the highest level observed during the study period. The area decreased again by approximately 34% in 2023\u0026ndash;2024 and 38% in 2024\u0026ndash;2025. The observed variability showed little correlation with precipitation, suggesting that non-climatic factors are dominant, especially since over 80% of wheat plots in Lebanon are supplementary irrigated. The sharp decline in 2021\u0026ndash;2022 was driven by the severe financial crisis, which included the suspension of the government subsidy program (removing a secure marketing channel), soaring inflation, and a huge increase in production costs, particularly for fuel and fertilizers. This made wheat cultivation a losing scenario, particularly for small farmers. The subsequent increase in 2022\u0026ndash;2023 coincided with a small degree of economic stabilization, including partial dollarization, which encouraged farmers to reinvest in wheat production. The decrease in 2023\u0026ndash;2024 and 2024\u0026ndash;2025 was linked to the renewed instability and insecurity caused by the Israeli offensive on Lebanon starting in October 2023, which impacted main wheat production areas and caused displacement. The high interannual variability is also consistent with common Lebanese agricultural practices, particularly crop rotation (e.g., wheat-potato rotation).\u003c/p\u003e \u003cp\u003eUsing the available data, we attempted to explain this variability, though further research is needed to fully understand it. While we analyzed the macro-economic factors across Lebanon along with precipitation trends, in a way to gain a general understanding of this variability, and to establish a baseline understanding of the situation. In order, to explain this variability in a holistic way, detailed information on the agriculture productivity and the agricultural practices of wheat farmers and their decision-making processes, across the different regions in Lebanon, are required. This would help identify the factors that generally influence farmers\u0026rsquo; crop choices, the changes that have occurred over time and their responses towards these changes; thereby, clarifying the reasons behind these patterns. Such information can be gathered through crop modeling for the productivity and socio-economic surveys with wheat farmers for the decision-making process. Consequently, future work should aim at deepening the analysis provided in this paper by addressing the wheat productivity over the cultivated areas and the practices of the farmers. Wheat productivity and yield will provide insights into the impact of fertilizers use, irrigation and pumping impacts, and access to fields during conflicts. Such exercise can be done by using the maps provided in this study as inputs to remote sensing driven crop modeling tools like Agri-Carbon- EO model (Al Bitar et al., 2022; Wijmer et al., 2024). In addition, agricultural practices and farmers decision-making will require adapted socio-economic surveys.\u003c/p\u003e"},{"header":"Declarations","content":"\u003col start=\"2\"\u003e\n \u003cli\u003e\u003cstrong\u003eAcknowledgment\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eThe authors gratefully acknowledge the financial support provided by the SAFAR program, which awarded a doctoral scholarship to Mariam Ibrahim; the PHC C\u0026egrave;dre program (project number 50473XL), which funded collaborative research activities between the French and Lebanese partners involved in this study; and the Water Diplomacy Center of Jordan (WDC) for providing a doctoral fellowship to Mariam Ibrahim.\u003cbr\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAbay, K. A., Breisinger, C., Glauber, J., Kurdi, S., Laborde, D., \u0026amp; Siddig, K. (2023). The Russia-Ukraine war: Implications for global and regional food security and potential policy responses. \u003cem\u003eGlobal Food Security\u003c/em\u003e, \u003cem\u003e36\u003c/em\u003e, 100675. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.gfs.2023.100675\u003c/span\u003e\u003cspan address=\"10.1016/j.gfs.2023.100675\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbdallah, A. (2025, March 6). \u003cem\u003eMeeting with Abbas Abdallah\u003c/em\u003e [Personal communication].\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAbou Ltaif, S. F., Mihai-Yiannaki, S., Thrassou, A., Ltaif, S. F. A., Mihai-Yiannaki, S., \u0026amp; Thrassou, A. (2024). Lebanon\u0026rsquo;s Economic Development Risk: Global Factors and Local Realities of the Shadow Economy Amid Financial Crisis. \u003cem\u003eRisks\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(8). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/risks12080122\u003c/span\u003e\u003cspan address=\"10.3390/risks12080122\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAcevedo, E., Silva, P., \u0026amp; Silva, H. (2002). Wheat growth and physiology. In \u003cem\u003eBread Wheat Improvement and Production\u003c/em\u003e (pp. 39\u0026ndash;70). FAO.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAchli, S., Ongoma, V., Epule, T. E., Dhiba, D., Salih, W., Ousayd, L., \u0026amp; Chehbouni, A. (2025). Exploring the key drivers of crop yields in Morocco \u0026ndash; a systematic review. \u003cem\u003eFrontiers in Agronomy\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fagro.2025.1515938\u003c/span\u003e\u003cspan address=\"10.3389/fagro.2025.1515938\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAitkenhead, M. J., \u0026amp; Dyer, R. (2007). Improving Land-cover Classification Using Recognition Threshold Neural Networks. \u003cem\u003ePhotogrammetric Engineering \u0026amp; Remote Sensing\u003c/em\u003e, \u003cem\u003e73\u003c/em\u003e(4), 413\u0026ndash;421. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.14358/PERS.73.4.413\u003c/span\u003e\u003cspan address=\"10.14358/PERS.73.4.413\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAl Bitar, A., Wijmer, T., Arnaud, L., Fieuzal, R., J-F, S., Gibrin, H., Ferlicop, M., J-F, D., \u0026amp; Ceschia, E. (2022). Quantification of the impact of cover crops on Net Ecosystem Exchange using AgriCarbon-EOv0.1. \u003cem\u003eIGARSS 2022\u0026ndash;2022 IEEE International Geoscience and Remote Sensing Symposium\u003c/em\u003e, 5781\u0026ndash;5784. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1109/IGARSS46834.2022.9883585\u003c/span\u003e\u003cspan address=\"10.1109/IGARSS46834.2022.9883585\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlami Machichi, M., El Mansouri, L., Imani, Y., Bourja, O., Hadria, R., Lahlou, O., Benmansour, S., Zennayi, Y., \u0026amp; Bourzeix, F. (2022). CerealNet: A Hybrid Deep Learning Architecture for Cereal Crop Mapping Using Sentinel-2 Time-Series. \u003cem\u003eInformatics\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(4), Article 4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/informatics9040096\u003c/span\u003e\u003cspan address=\"10.3390/informatics9040096\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAmare, M., Abay, K. A., Berhane, G., Andam, K. S., \u0026amp; Adeyanju, D. (2025). Conflicts, crop choice, and agricultural investments: Empirical evidence from Nigeria. \u003cem\u003eLand Use Policy\u003c/em\u003e, \u003cem\u003e148\u003c/em\u003e, 107391. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.landusepol.2024.107391\u003c/span\u003e\u003cspan address=\"10.1016/j.landusepol.2024.107391\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAntonenko, V., Al Bitar, A., Danylenko, I., Wijmer, T., Colin, J., Dejoux, J.-F., Lefebvre, A., Knibbe, M., Ceschia, E., \u0026amp; Gascoin, S. (2024). Impact of the Russian invasion on wheat biomass in Ukraine. \u003cem\u003eEnvironmental Research Letters\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1088/1748-9326/ad8363\u003c/span\u003e\u003cspan address=\"10.1088/1748-9326/ad8363\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArafeh, N., \u0026amp; Sukarieh, M. (2023). \u003cem\u003eBreaking the Cycle: Toward a New Imaginary of the Food System in Lebanon\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://carnegieendowment.org/research/2023/07/breaking-the-cycle-toward-a-new-imaginary-of-the-food-system-in-lebanon?center=middle-east\u0026amp;lang=en\u003c/span\u003e\u003cspan address=\"https://carnegieendowment.org/research/2023/07/breaking-the-cycle-toward-a-new-imaginary-of-the-food-system-in-lebanon?center=middle-east\u0026amp;lang=en\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBashour, I., Shehada Al-Ouda, A., Kassam, A., Bachour, R., Jouni, K., Hansmann, B., \u0026amp; Estephan, C. (2016). An overview of Conservation Agriculture in the dry Mediterranean environments with a special focus on Syria and Lebanon. \u003cem\u003eAIMS Agriculture and Food\u003c/em\u003e, \u003cem\u003e1\u003c/em\u003e, 67\u0026ndash;84. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3934/agrfood.2016.1.67\u003c/span\u003e\u003cspan address=\"10.3934/agrfood.2016.1.67\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerio Fortini, L., Chen, Q., Uyehara, Y., Bogner, K., Sprague, J., \u0026amp; Sprague, R. (2024). Fine-resolution land cover mapping over large and mountainous areas for Lāna\u0026lsquo;i, Hawaii using posterior probabilities, and expert knowledge. \u003cem\u003eInternational Journal of Remote Sensing\u003c/em\u003e, \u003cem\u003e45\u003c/em\u003e(6), 1949\u0026ndash;1971. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431161.2024.2321465\u003c/span\u003e\u003cspan address=\"10.1080/01431161.2024.2321465\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhattarai, G. P., Schmid, R. B., \u0026amp; McCornack, B. P. (2019). Remote Sensing Data to Detect Hessian Fly Infestation in Commercial Wheat Fields. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), Article 1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-019-42620-0\u003c/span\u003e\u003cspan address=\"10.1038/s41598-019-42620-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoaretto, A. E., Spolidoirio, E. S., Trivelin, P. C. O., Muraoka, T., Freitas, J. G. de, \u0026amp; Cantarella, H. (2000). Optimizing nitrogen-fertilizer application to wheat under irrigation. In \u003cem\u003eOptimizing nitrogen fertilizer application to irrigated wheat. Results of a co-ordinated research project. 1994\u0026ndash;1998\u003c/em\u003e (IAEA-TECDOC\u0026ndash;1164; pp. 63\u0026ndash;72). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://inis.iaea.org/records/rycfc-er588\u003c/span\u003e\u003cspan address=\"https://inis.iaea.org/records/rycfc-er588\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraga, P., Crusiol, L. G. T., Nanni, M. R., Caranhato, A. L. H., Fuhrmann, M. B., Nepomuceno, A. L., Neumaier, N., Farias, J. R. B., Koltun, A., Gon\u0026ccedil;alves, L. S. A., \u0026amp; Mertz-Henning, L. M. (2021). Vegetation indices and NIR-SWIR spectral bands as a phenotyping tool for water status determination in soybean. \u003cem\u003ePrecision Agriculture\u003c/em\u003e, \u003cem\u003e22\u003c/em\u003e(1), 249\u0026ndash;266. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/s11119-020-09740-4\u003c/span\u003e\u003cspan address=\"10.1007/s11119-020-09740-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman, L. (1996). Bagging predictors. \u003cem\u003eMachine Learning\u003c/em\u003e, \u003cem\u003e24\u003c/em\u003e(2), 123\u0026ndash;140. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1007/BF00058655\u003c/span\u003e\u003cspan address=\"10.1007/BF00058655\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreiman, L. (2001). Random Forests. \u003cem\u003eMachine Learning\u003c/em\u003e, \u003cem\u003e45\u003c/em\u003e(1), 5\u0026ndash;32. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1023/A:1010933404324\u003c/span\u003e\u003cspan address=\"10.1023/A:1010933404324\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreisinger, C., Khouri, N., Glauber, J. W., \u0026amp; Laborde Debucquet, D. (2023). \u003cem\u003eOne of the world\u0026rsquo;s worst economic collapses, now compounded by the Ukraine crisis: What\u0026rsquo;s next for Lebanon?\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hdl.handle.net/10568/140097\u003c/span\u003e\u003cspan address=\"https://hdl.handle.net/10568/140097\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBwambale, E., Naangmenyele, Z., Iradukunda, P., Agboka, K. M., Houessou-Dossou, E. A. Y., Akansake, D. A., Bisa, M. E., Hamadou, A.-A. H., Hakizayezu, J., Onofua, O. E., \u0026amp; Chikabvumbwa, S. R. (2022). Towards precision irrigation management: A review of GIS, remote sensing and emerging technologies. \u003cem\u003eCogent Engineering\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(1), 2100573. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/23311916.2022.2100573\u003c/span\u003e\u003cspan address=\"10.1080/23311916.2022.2100573\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaiserman, A. (2020). \u003cem\u003eAdapter les strat\u0026eacute;gies agricoles aux \u0026eacute;volutions socio-\u0026eacute;conomiques et climatiques en milieu m\u0026eacute;diterran\u0026eacute;en: Comparaison de l\u0026rsquo;usage de l\u0026rsquo;eau et des choix de plantes des agriculteurs dans la plaine de la B\u0026eacute;kaa (Liban) et de Marvdasht (Iran)\u003c/em\u003e [These de doctorat, Lyon]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.theses.fr/2020LYSE3009\u003c/span\u003e\u003cspan address=\"https://www.theses.fr/2020LYSE3009\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaiserman, A., Dumas, D., Bennafla, K., Faour, G., \u0026amp; Amiraslani, F. (2019). Application of Remotely Sensed Imagery and Socioeconomic Surveys to Map Crop Choices in the Bekaa Valley (Lebanon). \u003cem\u003eAgriculture\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(3), Article 3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/agriculture9030057\u003c/span\u003e\u003cspan address=\"10.3390/agriculture9030057\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCeccato, P., Gobron, N., Flasse, S., Pinty, B., \u0026amp; Tarantola, S. (2002). Designing a spectral index to estimate vegetation water content from remote sensing data: Part 1: Theoretical approach. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e82\u003c/em\u003e(2), 188\u0026ndash;197. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0034-4257(02)00037-8\u003c/span\u003e\u003cspan address=\"10.1016/S0034-4257(02)00037-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChaves, M. E. D., Picoli, M. C. A., \u0026amp; Sanches, I. D. (2020). Recent Applications of Landsat 8/OLI and Sentinel-2/MSI for Land Use and Land Cover Mapping: A Systematic Review. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(18), Article 18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs12183062\u003c/span\u003e\u003cspan address=\"10.3390/rs12183062\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCNRS-L, \u0026amp; UNDP Lebanon. (2024). \u003cem\u003eIsraeli Offensive on Lebanon 2023\u0026ndash;2024: Overview of attacks and damages across key sectors\u003c/em\u003e (p. 58). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cnrs.edu.lb/Library/Files/Uploaded%20Files/CNRS-L%20report%20on%20Israeli%20Offensive%20against%20Lebanon%202023-2024%20English.pdf\u003c/span\u003e\u003cspan address=\"https://www.cnrs.edu.lb/Library/Files/Uploaded%20Files/CNRS-L%20report%20on%20Israeli%20Offensive%20against%20Lebanon%202023-2024%20English.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCongalton, R. G., \u0026amp; Green, K. (2019). \u003cem\u003eAssessing the Accuracy of Remotely Sensed Data: Principles and Practices, Third Edition\u003c/em\u003e (3rd ed.). CRC Press. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1201/9780429052729\u003c/span\u003e\u003cspan address=\"10.1201/9780429052729\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDal, E., D\u0026iacute;az-Gonz\u0026aacute;lez, A. M., Morales-Opazo, C., \u0026amp; Vigani, M. (2021). \u003cem\u003eAgricultural sector review in Lebanon\u003c/em\u003e (FAO Agricultural Development Economics Technical Study No. 12). FAO; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openknowledge.fao.org/handle/20.\u003c/span\u003e\u003cspan address=\"https://openknowledge.fao.org/handle/20.\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e500.14283/cb5157en\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDarwish, T., Zdruli, P., Saliba, R., Awad, M., Shaban, A., \u0026amp; Faour, G. (2012). Vulnerability to Desertification in Lebanon Based on Geo-information and Socioeconomic Conditions. \u003cem\u003eJournal of Environmental Science and Engineering B\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDelegido, J., Verrelst, J., Alonso, L., \u0026amp; Moreno, J. (2011). Evaluation of Sentinel-2 Red-Edge Bands for Empirical Estimation of Green LAI and Chlorophyll Content. \u003cem\u003eSensors\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(7), Article 7. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/s110707063\u003c/span\u003e\u003cspan address=\"10.3390/s110707063\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDevadoss, S., \u0026amp; Ridley, W. (2024). Impacts of the Russian invasion of Ukraine on the global wheat market. \u003cem\u003eWorld Development\u003c/em\u003e, \u003cem\u003e173\u003c/em\u003e, 106396. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.worlddev.2023.106396\u003c/span\u003e\u003cspan address=\"10.1016/j.worlddev.2023.106396\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEuropean Space Agency. (2021). \u003cem\u003eCopernicus Sentinel-2: MSI Level-2A BOA Reflectance Product. Collection 1\u003c/em\u003e [Dataset]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5270/S2_-znk9xsj\u003c/span\u003e\u003cspan address=\"10.5270/S2_-znk9xsj\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO. (2006). \u003cem\u003eFood Security\u003c/em\u003e [Policy Brief]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.fao.org/fileadmin/templates/faoitaly/documents/pdf/pdf_Food_Security_Cocept_Note.pdf\u003c/span\u003e\u003cspan address=\"https://www.fao.org/fileadmin/templates/faoitaly/documents/pdf/pdf_Food_Security_Cocept_Note.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO. (2020). \u003cem\u003eSpecial Report \u0026ndash; FAO Mission to Assess the Impact of the Financial Crisis on Agriculture in the Republic of Lebanon\u003c/em\u003e. FAO. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4060/cb1164en\u003c/span\u003e\u003cspan address=\"10.4060/cb1164en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO. (2023a). \u003cem\u003eFood and Agriculture Organization Statistics\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.fao.org/faostat/en/#data/QCL\u003c/span\u003e\u003cspan address=\"https://www.fao.org/faostat/en/#data/QCL\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO GIEWS. (2018, January 28). \u003cem\u003eGlobal information and early warning system on food and agriculture (GIEWS) Country Brief: Lebanon 23-January-2018\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://reliefweb.int/report/lebanon/giews-country-brief-lebanon-23-january-2018\u003c/span\u003e\u003cspan address=\"https://reliefweb.int/report/lebanon/giews-country-brief-lebanon-23-january-2018\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO (with IFAD, UNICEF, WFP, \u0026amp; WHO). (2023b). \u003cem\u003eThe State of Food Security and Nutrition in the World 2023: Urbanization, agrifood systems transformation and healthy diets across the rural\u0026ndash;urban continuum\u003c/em\u003e. FAO. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4060/cc3017en\u003c/span\u003e\u003cspan address=\"10.4060/cc3017en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFAO, L. and W. D. (1971). \u003cem\u003eA study on the response of wheat to fertilizers.\u003c/em\u003e (1st ed.). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://openknowledge.fao.org/handle/20.500\u003c/span\u003e\u003cspan address=\"https://openknowledge.fao.org/handle/20.500\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.14283/aq377e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFaour, G., \u0026amp; Abdallah, C. (2020). \u003cem\u003eMapping Lands Cultivated with Wheat or Barley for the Year 2020 using Remote Sensing Technology\u003c/em\u003e (p. 234) [Internal Report]. CNRS Lebanon.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFayad, A., Gascoin, S., Faour, G., Fanise, P., Drapeau, L., Somma, J., Fadel, A., Al Bitar, A., \u0026amp; Escadafal, R. (2017). Snow observations in Mount Lebanon (2011\u0026ndash;2016). \u003cem\u003eEarth System Science Data\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(2), 573\u0026ndash;587. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/essd-9-573-2017\u003c/span\u003e\u003cspan address=\"10.5194/essd-9-573-2017\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeller, W. (1991). \u003cem\u003eAn Introduction to Probability Theory and Its Applications, Volume 2\u003c/em\u003e. John Wiley \u0026amp; Sons.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFern\u0026aacute;ndez-Manso, A., Fern\u0026aacute;ndez-Manso, O., \u0026amp; Quintano, C. (2016). SENTINEL-2A red-edge spectral indices suitability for discriminating burn severity. \u003cem\u003eInternational Journal of Applied Earth Observation and Geoinformation\u003c/em\u003e, \u003cem\u003e50\u003c/em\u003e, 170\u0026ndash;175. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jag.2016.03.005\u003c/span\u003e\u003cspan address=\"10.1016/j.jag.2016.03.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFoody, G. M. (2002). Status of land cover classification accuracy assessment. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e80\u003c/em\u003e(1), 185\u0026ndash;201. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0034-4257(01)00295-4\u003c/span\u003e\u003cspan address=\"10.1016/S0034-4257(01)00295-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGallego, J., Carfagna, E., \u0026amp; Baruth, B. (2010). Accuracy, Objectivity and Efficiency of Remote Sensing for Agricultural Statistics. In \u003cem\u003eAgricultural Survey Methods\u003c/em\u003e (pp. 193\u0026ndash;211). John Wiley \u0026amp; Sons, Ltd. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1002/9780470665480.ch12\u003c/span\u003e\u003cspan address=\"10.1002/9780470665480.ch12\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiri, C., Pengra, B., Long, J., \u0026amp; Loveland, T. R. (2013). Next generation of global land cover characterization, mapping, and monitoring. \u003cem\u003eInternational Journal of Applied Earth Observation and Geoinformation\u003c/em\u003e, \u003cem\u003e25\u003c/em\u003e, 30\u0026ndash;37. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jag.2013.03.005\u003c/span\u003e\u003cspan address=\"10.1016/j.jag.2013.03.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGoogle for Developers. (2025). \u003cem\u003eGoogle Earth Engine\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://developers.google.com/earth-engine\u003c/span\u003e\u003cspan address=\"https://developers.google.com/earth-engine\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGordon, S. I. (1980). Utilizing LANDSAT imagery to monitor land-use change: A case study in ohio. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(3), 189\u0026ndash;196. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0034-4257(80)90028-0\u003c/span\u003e\u003cspan address=\"10.1016/0034-4257(80)90028-0\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHall, D. K., Riggs, G. A., \u0026amp; Salomonson, V. V. (1995). Development of methods for mapping global snow cover using moderate resolution imaging spectroradiometer data. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e, 127\u0026ndash;140. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0034-4257(95)00137-P\u003c/span\u003e\u003cspan address=\"10.1016/0034-4257(95)00137-P\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHashmiu, I., Agbenyega, O., \u0026amp; Dawoe, E. (2022). Determinants of crop choice decisions under risk: A case study on the revival of cocoa farming in the Forest-Savannah transition zone of Ghana. \u003cem\u003eLand Use Policy\u003c/em\u003e, \u003cem\u003e114\u003c/em\u003e, 105958. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.landusepol.2021.105958\u003c/span\u003e\u003cspan address=\"10.1016/j.landusepol.2021.105958\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHowarth, J. P., \u0026amp; Wickware, G. M. (1981). Procedures for change detection using Landsat digital data. \u003cem\u003eInternational Journal of Remote Sensing\u003c/em\u003e, \u003cem\u003e2\u003c/em\u003e(3), 277\u0026ndash;291.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHunt, E. R., \u0026amp; Rock, B. N. (1989). Detection of changes in leaf water content using Near- and Middle-Infrared reflectances. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e30\u003c/em\u003e(1), 43\u0026ndash;54. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0034-4257(89)90046-1\u003c/span\u003e\u003cspan address=\"10.1016/0034-4257(89)90046-1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eImmitzer, M., Vuolo, F., \u0026amp; Atzberger, C. (2016). First Experience with Sentinel-2 Data for Crop and Tree Species Classifications in Central Europe. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(3), Article 3. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs8030166\u003c/span\u003e\u003cspan address=\"10.3390/rs8030166\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIPCC. (2023). Summary for Policymakers. In \u003cem\u003eClimate Change 2023: Synthesis Report. Contribution of Working Groups I, II and III to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change\u003c/em\u003e ([Core Writing Team, H. Lee and J. Romero (eds.)]., pp. 1\u0026ndash;34).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIPT Group. (2025). \u003cem\u003eFuel Prices\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.iptgroup.com.lb/Sub.aspx?pageid=22\u003c/span\u003e\u003cspan address=\"https://www.iptgroup.com.lb/Sub.aspx?pageid=22\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJaafar, H., Karimi, P., \u0026amp; Borgomeo, E. (2024). Economic irrigation water productivity of wheat and potato: An earth observation perspective on policy implications in the Litani Basin, Lebanon. \u003cem\u003eAgricultural Water Management\u003c/em\u003e, \u003cem\u003e306\u003c/em\u003e, 109180. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.agwat.2024.109180\u003c/span\u003e\u003cspan address=\"10.1016/j.agwat.2024.109180\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhabba, S., Er-Raki, S., Toumi, J., Ezzahar, J., Ait Hssaine, B., Le Page, M., \u0026amp; Chehbouni, A. (2020). A Simple Light-Use-Efficiency Model to Estimate Wheat Yield in the Semi-Arid Areas. \u003cem\u003eAgronomy\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(10), Article 10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/agronomy10101524\u003c/span\u003e\u003cspan address=\"10.3390/agronomy10101524\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhafagy, A., D\u0026iacute;az-Gonz\u0026aacute;lez, A. M., Nano, E., Soguero Escuer, J., Morales Opazo, C., Salibi, A., El Tawn, L., \u0026amp; Dikah, W. (2022). \u003cem\u003eLebanon\u0026rsquo;s agrifood system in times of turbulence: Obstacles and opportunities\u003c/em\u003e. FAO. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.4060/cc2334en\u003c/span\u003e\u003cspan address=\"10.4060/cc2334en\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhan, A., Ahmad, M., Ahmed, M., \u0026amp; Iftikhar Hussain, M. (2020). Rising Atmospheric Temperature Impact on Wheat and Thermotolerance Strategies. \u003cem\u003ePlants\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(1), 43. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/plants10010043\u003c/span\u003e\u003cspan address=\"10.3390/plants10010043\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhlif, M., Escorihuela, M. J., Chahbi Bellakanji, A., Paolini, G., Kassouk, Z., \u0026amp; Lili Chabaane, Z. (2023). Multi-Year Cereal Crop Classification Model in a Semi-Arid Region Using Sentinel-2 and Landsat 7\u0026ndash;8 Data. \u003cem\u003eAgriculture\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(8), Article 8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/agriculture13081633\u003c/span\u003e\u003cspan address=\"10.3390/agriculture13081633\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKoweyes, J., Salloum, T., Haidar, S., Merhi, G., \u0026amp; Tokajian, S. (2021). COVID-19 Pandemic in Lebanon: One Year Later, What Have We Learnt? \u003cem\u003emSystems\u003c/em\u003e, \u003cem\u003e6\u003c/em\u003e(2), e00351-21. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1128/mSystems.00351-21\u003c/span\u003e\u003cspan address=\"10.1128/mSystems.00351-21\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuhla, K., Puma, M. J., \u0026amp; Otto, C. (2024). International cooperation was key to stabilize wheat prices after the Russian Invasion of Ukraine. \u003cem\u003eCommunications Earth \u0026amp; Environment\u003c/em\u003e, \u003cem\u003e5\u003c/em\u003e(1), 481. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s43247-024-01638-7\u003c/span\u003e\u003cspan address=\"10.1038/s43247-024-01638-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLago-Olveira, S., Ouhemi, H., Idrissi, O., Moreira, M. T., \u0026amp; Gonz\u0026aacute;lez-Garc\u0026iacute;a, S. (2024). Promoting more sustainable agriculture in the Moroccan drylands by shifting from conventional wheat monoculture to a rotation with chickpea and lentils. \u003cem\u003eCleaner Environmental Systems\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e, 100169. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.cesys.2024.100169\u003c/span\u003e\u003cspan address=\"10.1016/j.cesys.2024.100169\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLandis, J. R., \u0026amp; Koch, G. G. (1977). The Measurement of Observer Agreement for Categorical Data. \u003cem\u003eBiometrics\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(1), 159\u0026ndash;174. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/2529310\u003c/span\u003e\u003cspan address=\"10.2307/2529310\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLe, T. H. L., Kristiansen, P., Vo, B., Moss, J., \u0026amp; Welch, M. (2024). Understanding factors influencing farmers\u0026rsquo; crop choice and agricultural transformation in the Upper Vietnamese Mekong Delta. \u003cem\u003eAgricultural Systems\u003c/em\u003e, \u003cem\u003e216\u003c/em\u003e, 103899. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.agsy.2024.103899\u003c/span\u003e\u003cspan address=\"10.1016/j.agsy.2024.103899\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLebanese Customs. (2024). \u003cem\u003eTrade Statistics\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.customs.gov.lb/Trade_Statistics/Yearly/Yearly_Statistics.aspx\u003c/span\u003e\u003cspan address=\"http://www.customs.gov.lb/Trade_Statistics/Yearly/Yearly_Statistics.aspx\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLidon, F. C., Almeida, A. S., Leitao, A. L. cia, Silva, M. M., Pinheiro, N., Macas, B., \u0026amp; Costa, R. (2014). A SYNOPTIC OVERVIEW OF DURUM WHEAT PRODUCTION IN THE MEDITERRANEAN REGION AND PROCESSING FOLLOWING THE EUROPEAN UNION REQUIREMENTS. \u003cem\u003eEmirates Journal of Food and Agriculture\u003c/em\u003e, 693\u0026ndash;705. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.9755/ejfa.v26i8.17066\u003c/span\u003e\u003cspan address=\"10.9755/ejfa.v26i8.17066\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, J., Feng, Q., Gong, J., Zhou, J., Liang, J., \u0026amp; Li, Y. (2018). Winter wheat mapping using a random forest classifier combined with multi-temporal and multi-sensor data. \u003cem\u003eInternational Journal of Digital Earth\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(8), 783\u0026ndash;802. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/17538947.2017.1356388\u003c/span\u003e\u003cspan address=\"10.1080/17538947.2017.1356388\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMacintyre, P., van Niekerk, A., \u0026amp; Mucina, L. (2020). Efficacy of multi-season Sentinel-2 imagery for compositional vegetation classification. \u003cem\u003eInternational Journal of Applied Earth Observation and Geoinformation\u003c/em\u003e, \u003cem\u003e85\u003c/em\u003e, 101980. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.jag.2019.101980\u003c/span\u003e\u003cspan address=\"10.1016/j.jag.2019.101980\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMaddah, L., \u0026amp; Darwish, S. (2023). \u003cem\u003eThe Road to Recovery for Lebanese Agriculture\u003c/em\u003e. The Lebanese Center for Policy Studies. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.lcps-lebanon.org/en/articles/details/4821/the-road-to-recovery-for-lebanese-agriculture\u003c/span\u003e\u003cspan address=\"https://www.lcps-lebanon.org/en/articles/details/4821/the-road-to-recovery-for-lebanese-agriculture\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMain-Knorn, M., Pflug, B., Louis, J., Debaecker, V., M\u0026uuml;ller-Wilm, U., \u0026amp; Gascon, F. (2017). Sen2Cor for Sentinel-2. \u003cem\u003eImage and Signal Processing for Remote Sensing XXIII\u003c/em\u003e, \u003cem\u003e10427\u003c/em\u003e, 37\u0026ndash;48. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1117/12.2278218\u003c/span\u003e\u003cspan address=\"10.1117/12.2278218\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMakdisi, S., \u0026amp; Amine, R. (2022). Lebanon\u0026rsquo;s Multifaceted Economic Crisis of October 2019: Causes, Repercussions- A Diagnosis. \u003cem\u003eEconomic Research Forum\u003c/em\u003e, \u003cem\u003e1562\u003c/em\u003e, Article 1562. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://ideas.repec.org//p/erg/wpaper/1562.html\u003c/span\u003e\u003cspan address=\"https://ideas.repec.org//p/erg/wpaper/1562.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcFEETERS, S. K. (1996). The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. \u003cem\u003eInternational Journal of Remote Sensing\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(7), 1425\u0026ndash;1432. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431169608948714\u003c/span\u003e\u003cspan address=\"10.1080/01431169608948714\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Agriculture, \u0026amp; FAO. (2010). \u003cem\u003eالإحصاء الزراعي الشامل Comprehensive Agricultural Census\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.agriculture.gov.lb/Statistics-and-Studies/Comprehensive-Agricultural-Statistics\u003c/span\u003e\u003cspan address=\"http://www.agriculture.gov.lb/Statistics-and-Studies/Comprehensive-Agricultural-Statistics\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMinistry of Finance. (2012). \u003cem\u003eLebanon\u0026mdash;Wheat and Bread Subsidy Policies 2007\u0026ndash;2011\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.arabdevelopmentportal.com/publication/lebanon-wheat-and-bread-subsidy-policies-2007-2011\u003c/span\u003e\u003cspan address=\"https://www.arabdevelopmentportal.com/publication/lebanon-wheat-and-bread-subsidy-policies-2007-2011\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMyneni, R. B., \u0026amp; Williams, D. L. (1994). On the relationship between FAPAR and NDVI. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e49\u003c/em\u003e(3), 200\u0026ndash;211. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0034-4257(94)90016-7\u003c/span\u003e\u003cspan address=\"10.1016/0034-4257(94)90016-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNarin, O. G., Abdikan, S., Bayik, C., Sekertekin, A., Delen, A., \u0026amp; Balik Sanli, F. (2021). COHERENCE and BACKSCATTER BASED CROPLAND MAPPING USING MULTI-TEMPORAL SENTINEL-1 with DYNAMIC TIME WARPING. \u003cem\u003eThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences\u003c/em\u003e, \u003cem\u003eXLIII-B5-2021\u003c/em\u003e, 37\u0026ndash;41. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/isprs-archives-XLIII-B5-2021-37-2021\u003c/span\u003e\u003cspan address=\"10.5194/isprs-archives-XLIII-B5-2021-37-2021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNASA. (2023). \u003cem\u003eTiling System\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://hls.gsfc.nasa.gov/products-description/tiling-system/\u003c/span\u003e\u003cspan address=\"https://hls.gsfc.nasa.gov/products-description/tiling-system/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNasrallah, A. (2019). \u003cem\u003eCrop mapping and yield estimation of wheat in the Bekaa plain of Lebanon\u003c/em\u003e [Phdthesis, AgroParisTech]. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pastel.hal.science/tel-02520013\u003c/span\u003e\u003cspan address=\"https://pastel.hal.science/tel-02520013\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNasrallah, A., Baghdadi, N., El Hajj, M., Darwish, T., Belhouchette, H., Faour, G., Darwich, S., \u0026amp; Mhawej, M. (2019). Sentinel-1 Data for Winter Wheat Phenology Monitoring and Mapping. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e11\u003c/em\u003e(19), Article 19. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs11192228\u003c/span\u003e\u003cspan address=\"10.3390/rs11192228\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNasrallah, A., Baghdadi, N., Mhawej, M., Faour, G., Darwish, T., Belhouchette, H., \u0026amp; Darwich, S. (2018). A Novel Approach for Mapping Wheat Areas Using High Resolution Sentinel-2 Images. \u003cem\u003eSensors (Basel, Switzerland)\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(7), 2089. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/s18072089\u003c/span\u003e\u003cspan address=\"10.3390/s18072089\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOCHA, R. (2024, December 23). \u003cem\u003eLebanon at War Crisis Update, November 2024\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://reliefweb.int/report/lebanon/lebanon-war-crisis-update-november-2024\u003c/span\u003e\u003cspan address=\"https://reliefweb.int/report/lebanon/lebanon-war-crisis-update-november-2024\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh, S., Chang, A., Ashapure, A., Jung, J., Dube, N., Maeda, M., Gonzalez, D., \u0026amp; Landivar, J. (2020). Plant Counting of Cotton from UAS Imagery Using Deep Learning-Based Object Detection Framework. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(18), Article 18. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs12182981\u003c/span\u003e\u003cspan address=\"10.3390/rs12182981\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlofsson, P., Ar\u0026eacute;valo, P., Espejo, A. B., Green, C., Lindquist, E., McRoberts, R. E., \u0026amp; Sanz, M. J. (2020). Mitigating the effects of omission errors on area and area change estimates. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e236\u003c/em\u003e, 111492. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rse.2019.111492\u003c/span\u003e\u003cspan address=\"10.1016/j.rse.2019.111492\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlofsson, P., Foody, G. M., Herold, M., Stehman, S. V., Woodcock, C. E., \u0026amp; Wulder, M. A. (2014). Good practices for estimating area and assessing accuracy of land change. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e148\u003c/em\u003e, 42\u0026ndash;57. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rse.2014.02.015\u003c/span\u003e\u003cspan address=\"10.1016/j.rse.2014.02.015\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlofsson, P., Foody, G. M., Stehman, S. V., \u0026amp; Woodcock, C. E. (2013). Making better use of accuracy data in land change studies: Estimating accuracy and area and quantifying uncertainty using stratified estimation. \u003cem\u003eRemote Sensing of Environment\u003c/em\u003e, \u003cem\u003e129\u003c/em\u003e, 122\u0026ndash;131. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rse.2012.10.031\u003c/span\u003e\u003cspan address=\"10.1016/j.rse.2012.10.031\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eProbst, P., \u0026amp; Boulesteix, A.-L. (2018). To Tune or Not to Tune the Number of Trees in Random Forest. \u003cem\u003eJournal of Machine Learning Research\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(181), 1\u0026ndash;18.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRodriguez-Galiano, V. F., Ghimire, B., Rogan, J., Chica-Olmo, M., \u0026amp; Rigol-Sanchez, J. P. (2012). An assessment of the effectiveness of a random forest classifier for land-cover classification. \u003cem\u003eISPRS Journal of Photogrammetry and Remote Sensing\u003c/em\u003e, \u003cem\u003e67\u003c/em\u003e, 93\u0026ndash;104. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.isprsjprs.2011.11.002\u003c/span\u003e\u003cspan address=\"10.1016/j.isprsjprs.2011.11.002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRouse, J., Haas, R. H., Schell, J. A., \u0026amp; Deering, D. (1973). \u003cem\u003eMonitoring vegetation systems in the great plains with ERTS\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.semanticscholar.org/paper/Monitoring-vegetation-systems-in-the-great-plains-Rouse-Haas/fb2f60fe0fe2874e\u003c/span\u003e\u003cspan address=\"https://www.semanticscholar.org/paper/Monitoring-vegetation-systems-in-the-great-plains-Rouse-Haas/fb2f60fe0fe2874e\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e5cbf927a2556d719c32eac29\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSafa, M., Samarasinghe, S., \u0026amp; Mohssen, M. (2010). Determination of fuel consumption and indirect factors affecting it in wheat production in Canterbury, New Zealand. \u003cem\u003eEnergy\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e(12), 5400\u0026ndash;5405. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.energy.2010.07.015\u003c/span\u003e\u003cspan address=\"10.1016/j.energy.2010.07.015\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanlaville, P. (1963). \u003cem\u003eLes r\u0026eacute;gions agricoles du Liban\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3406/geoca.1963.1751\u003c/span\u003e\u003cspan address=\"10.3406/geoca.1963.1751\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScience Toolbox Exploitation Platform, E. (2024). Sen2Cor. \u003cem\u003eSNAP Supported Plugins\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://step.esa.int/main/snap-supported-plugins/sen2cor/\u003c/span\u003e\u003cspan address=\"https://step.esa.int/main/snap-supported-plugins/sen2cor/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSINGH, A. (1989). Review Article Digital change detection techniques using remotely-sensed data. \u003cem\u003eInternational Journal of Remote Sensing\u003c/em\u003e, \u003cem\u003e10\u003c/em\u003e(6), 989\u0026ndash;1003. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431168908903939\u003c/span\u003e\u003cspan address=\"10.1080/01431168908903939\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSothe, C., Almeida, C. M. de, Liesenberg, V., \u0026amp; Schimalski, M. B. (2017). Evaluating Sentinel-2 and Landsat-8 Data to Map Sucessional Forest Stages in a Subtropical Forest in Southern Brazil. \u003cem\u003eRemote Sensing\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(8), Article 8. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3390/rs9080838\u003c/span\u003e\u003cspan address=\"10.3390/rs9080838\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTadesse, W., Amri, A., Sanchez-Garcia, M., El-Bouhssini, M., Karrou, M., \u0026amp; Patil, S. (2017). Improving wheat production in the Central and West Asia and North Africa (CWANA) region. In \u003cem\u003eAchieving sustainable cultivation of wheat\u003c/em\u003e (Vol. 2). Burleigh Dodds Science Publishing.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTalhouk, S. N., Zurayk, R., \u0026amp; Khuri, S. (2001). Conservation of the coniferous forests of Lebanon: Past, present and future prospects. \u003cem\u003eOryx\u003c/em\u003e, \u003cem\u003e35\u003c/em\u003e(3), 206\u0026ndash;215. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1046/j.1365-3008.2001.00180.x\u003c/span\u003e\u003cspan address=\"10.1046/j.1365-3008.2001.00180.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTawk, S. T., Chedid, M., Chalak, A., Karam, S., \u0026amp; Hamadeh, S. K. (2019). Challenges and Sustainability of Wheat Production in a Levantine Breadbasket: The Case of the West Bekaa, Lebanon. \u003cem\u003eJournal of Agriculture, Food Systems, and Community Development\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(4), Article 4. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5304/jafscd.2019.084.011\u003c/span\u003e\u003cspan address=\"10.5304/jafscd.2019.084.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTita, D., Mahdi, K., Devkota, K. P., \u0026amp; Devkota, M. (2025). Climate change and agronomic management: Addressing wheat yield gaps and sustainability challenges in the Mediterranean and MENA regions. \u003cem\u003eAgricultural Systems\u003c/em\u003e, \u003cem\u003e224\u003c/em\u003e, 104242. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.agsy.2024.104242\u003c/span\u003e\u003cspan address=\"10.1016/j.agsy.2024.104242\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTohme Tawk, S., Chedid, M., Chalak, A., Karam, S., \u0026amp; Hamadeh, S. (2019). Challenges and Sustainability of Wheat Production in a Levantine Breadbasket: The Case of the West Bekaa, Lebanon. \u003cem\u003eJournal of Agriculture, Food Systems, and Community Development\u003c/em\u003e, 1\u0026ndash;17. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5304/jafscd.2019.084.011\u003c/span\u003e\u003cspan address=\"10.5304/jafscd.2019.084.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUN Office for the Coordination of Humanitarian Affairs. (2020, August 7). \u003cem\u003eLebanon: Beirut Port Explosions Situation Report No. 2 (As of 7 August 2020)\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://reliefweb.int/report/lebanon/lebanon-beirut-port-explosions-situation-report-no-2-7-august-2020-enar\u003c/span\u003e\u003cspan address=\"https://reliefweb.int/report/lebanon/lebanon-beirut-port-explosions-situation-report-no-2-7-august-2020-enar\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUnited Nations. (2023). \u003cem\u003eLebanon Crisis Response Plan (LCRP)\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://data.unhcr.org/en/documents/details/100389\u003c/span\u003e\u003cspan address=\"https://data.unhcr.org/en/documents/details/100389\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVerdeil, E., Faour, G., \u0026amp; Hamze, M. (2016). Atlas du Liban: Les nouveaux d\u0026eacute;fis. In \u003cem\u003eAtlas du Liban: Les nouveaux d\u0026eacute;fis\u003c/em\u003e. Presses de l\u0026rsquo;Ifpo. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://books.openedition.org/ifpo/10709\u003c/span\u003e\u003cspan address=\"https://books.openedition.org/ifpo/10709\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVolsi, B., Higashi, G. E., Bordin, I., \u0026amp; Telles, T. S. (2022). The diversification of species in crop rotation increases the profitability of grain production systems. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 19849. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41598-022-23718-4\u003c/span\u003e\u003cspan address=\"10.1038/s41598-022-23718-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWagner, J., Skakun, S., Nair, S. S., Sadeh, Y., Baber, S., Oliinyk, O., Kotcharlakota,A., Gupta, M., Poliakov, D., Vaskivskyi, B., Misiura, O., Kussul, N., Prykhodko, D.,Sikachyna, O., Rajaoberison, A., Li, F., Chevassu, M., Rehbinder, J., Nerry, F., \u0026hellip;Becker-Reshef, I. (2026). Monitoring winter crop areas during wartime: Remote sensing support for Ukraine\u0026rsquo;s agricultural statistics. \u003cem\u003eNpj Sustainable Agriculture\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e(1), 1. https://doi.org/10.1038/s44264-025-00119-4\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, D., Li, Z., Zhou, Q., Chen, Z., \u0026amp; Liu, J. (2015). Estimation of winter wheat acreage via a combination of remotely sensed data and an optimized spatial sampling scheme. \u003cem\u003eInternational Journal of Remote Sensing\u003c/em\u003e, \u003cem\u003e36\u003c/em\u003e(19\u0026ndash;20), 5208\u0026ndash;5221. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1080/01431161.2015.1093197\u003c/span\u003e\u003cspan address=\"10.1080/01431161.2015.1093197\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWijmer, T., Al Bitar, A., Arnaud, L., Fieuzal, R., \u0026amp; Ceschia, E. (2024). AgriCarbon-EO v1.0.1: Large-scale and high-resolution simulation of carbon fluxes by assimilation of Sentinel-2 and Landsat-8 reflectances using a Bayesian approach. \u003cem\u003eGeoscientific Model Development\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(3), 997\u0026ndash;1021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.5194/gmd-17-997-2024\u003c/span\u003e\u003cspan address=\"10.5194/gmd-17-997-2024\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Bank Group. (2022). \u003cem\u003eLebanon\u0026mdash;Wheat Supply Emergency Response Project\u003c/em\u003e. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://documents.worldbank.org/curated/en/408131653327258940\u003c/span\u003e\u003cspan address=\"http://documents.worldbank.org/curated/en/408131653327258940\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, X., Xiong, J., Du, T., Ju, X., Gan, Y., Li, S., Xia, L., Shen, Y., Pacenka, S., Steenhuis, T. S., Siddique, K. H. M., Kang, S., \u0026amp; Butterbach-Bahl, K. (2024). Diversifying crop rotation increases food production, reduces net greenhouse gas emissions and improves soil health. \u003cem\u003eNature Communications\u003c/em\u003e, \u003cem\u003e15\u003c/em\u003e(1), Article 1. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1038/s41467-023-44464-9\u003c/span\u003e\u003cspan address=\"10.1038/s41467-023-44464-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, Z., Islam, F., Waseem, L. A., Tariq, A., Nawaz, M., Islam, I. U., Bibi, T., Rehman, N. U., Ahmad, W., Aslam, R. W., Raza, D., \u0026amp; Hatamleh, W. A. (2024). Comparison of Three Machine Learning Algorithms Using Google Earth Engine for Land Use Land Cover Classification. \u003cem\u003eRangeland Ecology \u0026amp; Management\u003c/em\u003e, \u003cem\u003e92\u003c/em\u003e, 129\u0026ndash;137. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rama.2023.10.007\u003c/span\u003e\u003cspan address=\"10.1016/j.rama.2023.10.007\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"Table 10","content":"\u003cp\u003eTable 10 is not available with this version.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"npj-sustainable-agriculture","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [npj Sustainable Agriculture](https://www.nature.com/npjsustainagric/)","snPcode":"44264","submissionUrl":"https://submission.springernature.com/new-submission/44264/3","title":"npj Sustainable Agriculture","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Wheat, Lebanon, Crisis, Crop Mapping, Sentinel-2, Random Forest, Food Security","lastPublishedDoi":"10.21203/rs.3.rs-8649264/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8649264/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWheat stands as one of the most important staple crops worldwide. However, the vital role of this crop has been increasingly challenged in Lebanon, in recent years by multi-factorial crises from socio-economic, political, security and climate factors, threatening agricultural stability and food supply. Consequently, monitoring wheat production is crucial for managing import and export activities, developing effective policies, achieving resilient agricultural development, and ensuring food security. This study provides the first national, multi-year monitoring of wheat area in Lebanon (2017\u0026ndash;2025), linking satellite observations with crisis impacts to support food security planning. We conducted a multi-temporal supervised classification from 2017\u0026ndash;2018 to 2024\u0026ndash;2025 seasons, using Google Earth Engine, employing Sentinel-2 optical images and Random Forest classifier. We estimated wheat area based on a random stratified sample achieving an overall accuracy of 87%. Interannual changes were then related to major crises and input-price dynamics. Wheat area increased during 2019\u0026ndash;2021 but dropped sharply in 2021\u0026ndash;2022 as subsidies weakened and input costs surged. Indeed, during the transition toward economy dollarization, the computed indicator of production cost expressed in USD peak in 2021\u0026ndash;2022 and then ease consistently with the 2022\u0026ndash;2023 area rebound reaching the highest level observed during the study period. In 2023\u0026ndash;2025, the crop area decreased again dramatically (-34% in 2023\u0026ndash;2024 and \u0026minus;\u0026thinsp;38% in 2024\u0026ndash;2025) in relation to the conflict with Israel and associated widespread displacement of population that likely constrained field access and reduced sowing, particularly in southern Lebanon. Our findings also point out the need of ground-truth data for accurate area estimation, as well as additional information on yield and socio-economic conditions to better understand the interannual variability of wheat area.\u003c/p\u003e","manuscriptTitle":"Wheat in Crisis: Variability of Wheat Area in Lebanon 2017-2025","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-03 15:33:53","doi":"10.21203/rs.3.rs-8649264/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"45025820151490974166221000713867111321","date":"2026-05-07T05:37:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"220051991598245665998890208423279611790","date":"2026-04-09T01:01:54+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-06T02:22:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"167398636106631537009421882768754673475","date":"2026-04-03T19:42:32+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"335553450414784381532360286552987681268","date":"2026-03-09T03:15:48+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"98340959933311350844011368959766486426","date":"2026-02-13T06:17:46+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-01-27T00:50:13+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-27T00:35:24+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-01-23T06:40:50+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Sustainable Agriculture","date":"2026-01-20T11:41:43+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-sustainable-agriculture","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [npj Sustainable Agriculture](https://www.nature.com/npjsustainagric/)","snPcode":"44264","submissionUrl":"https://submission.springernature.com/new-submission/44264/3","title":"npj Sustainable Agriculture","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"589abf7c-5ced-4aef-bac1-423192af19c2","owner":[],"postedDate":"February 3rd, 2026","published":true,"recentEditorialEvents":[{"type":"reviewerAgreed","content":"45025820151490974166221000713867111321","date":"2026-05-07T05:37:48+00:00","index":55,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":61785021,"name":"Earth and environmental sciences/Climate sciences"},{"id":61785022,"name":"Earth and environmental sciences/Environmental sciences"},{"id":61785023,"name":"Earth and environmental sciences/Environmental social sciences"}],"tags":[],"updatedAt":"2026-02-03T15:33:53+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-03 15:33:53","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8649264","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8649264","identity":"rs-8649264","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}