Spatio-temporal Monitoring of Drought using Machine Learning approach and Remote Sensing Techniques in Ningxia

Spatio-temporal Monitoring of Drought using Machine Learning approach and Remote Sensing Techniques in Ningxia | 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 Research Article Spatio-temporal Monitoring of Drought using Machine Learning approach and Remote Sensing Techniques in Ningxia Muhammad Awais Khan, Shawkat Ali, Zakria Zaheen, Hidayat Ullah, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5259358/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Timely and accurate monitoring of the beginning and development of drought in China is significant in decreasing losses from drought. The present study contributes to a comprehensive spatio-temporal analysis of drought over the Ningxia Hui (northwestern China) from 2003–2023. We determined the moisture content and vegetation using MODIS satellite data. The Enhanced Vegetation Index (EVI), the Land Surface Temperature (LST), the Standardized Precipitation Index (SPI-1, SPI-3, SPI-6, SPI-9 and SPI-12), and the Standardized Precipitation-Evapotranspiration Index (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12), were calculated. SPEI at 1–12 months timescales and the Keetch-Byram Drought Index (KBDI) were adopted to characterize drought events over the Ningxia region from 2003 to 2023. Future drought predictions were determined based on SPI at 1–12 months timescales using an artificial neural network (ANN) and cellular automata (CA) machine learning approaches. The CA-ANN model was used to validate drought prediction. The results showed: (1) the EVI declined from 0.38 to 0.33 from 2003–2023. This declining EVI indicates that the annual average of vegetation was decreased ; (2) The KBDI increased from 581.33 in 2003 to 681.091 in 2023, reflecting aggrading aridity with the soil moisture drying out; (3) SPI decreased from 0.7 in 2003 to -1.835 in 2023 and the SPEI varied from 0.5 to − 1.898 in the same period, (4) SPEI results in 2003 highlight western and southern parts highly affected by drought; (6) drought prediction from CA-ANN display that the SPI and SPEI expected in 2033 will further decrease and can cause more frequent drought. The study concluded that the ever-declining drought conditions in the Ningxia region over the past two decades have manifested drastic changes in the drought conditions. Drought Machine Learning EVI SPI SPEI KBDI ANN 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 Drought is an intricate and recurring natural disaster that influences agricultural production, water resources, ecology, and socio-economic development across scales from local communities to regions. Severe droughts can decimate crop yields, grazing lands, and livestock herds, stress water supplies, fuel wildfires, and lead to large economic losses [ 1 ]. Many regions of the world face more pronounced and, in some cases, more recurrent droughts as climate change changes historical drought patterns and their intensity. Understanding the historical variations of drought and its susceptibilities is imperative to facilitate adaptation and mitigation approaches [ 2 ]. Ningxia is in arid northwest China and has been prone to severe drought due to its desert climate with high temperatures and low annual precipitation. Foremost droughts have led to drinking and stock water deficiencies in recent decades and declined agricultural output in Ningxia and the surrounding areas [ 3 ]. In this study, we perform an inclusive spatio-temporal analysis of drought patterns in Ningxia from the last two decades (2003–2023). We combine remote sensing (MODIS) data and climate records of precipitation, temperature, evapotranspiration, and soil moisture to utilize MODIS constant exposure of vegetation health and drought stress across Ningxia from 2003 to 2023 and the temporal patterns in drought-related variables across the study area over the same period. Then, we use machine learning methods, such as random forests (RF) and regression trees (RTs), to model the correlations between the MODIS inputs and the severity of the drought throughout Ningxia [ 4 ]. Drought calamities, characterized by prolonged scarcities in precipitation, represent complex natural phenomena that considerably affect the environment, agriculture, and human society [ 5 ]. They are closely connected to the continuously changing climate system, and their occurrence and severity are deeply affected by variations in the atmosphere and persistent weather patterns. The combination of increasing temperatures and an ongoing decline in precipitation from long periods initiates a gradual reduction in soil moisture, lessening water supplies, and a loss of agricultural productivity. The vulnerabilities of regions already prone to dry conditions or relying on a scarce inventory of water resources are magnified with the arrival of drought, which can destroy food supplies, livelihoods, and freshwater availability. The modern period of climate change especially leads to a greater reappearance and severity of drought events across many regions of the world, with droughts becoming more recurrent and severe [ 6 ]. Understanding the complex interchange of drought and climate is of great scientific interest and is serious for effective disaster management and the creation of adaptable approaches. Such an understanding delivers the foundation for mitigating the multiplicity of effects connected with droughts, ensuring supportable agricultural techniques, and creating the resilience and health of both communities and ecosystems, as they face the enhancing challenge of these disastrous climate procedures [ 7 ]. Writing at the time of that original column, another team strained the need to nail down the definitions of drought, pointing out its individual and multidimensional nature [ 8 ]. They called for a holistic method that considered meteorological, hydrological, and socio-economic components to improve understanding of drought [ 9 ]. This multispectral perspective has become one of the fundamentals of drought research, delivering more detailed evaluations and the ability to more effectively craft mitigation plans [ 10 ]. Meanwhile, the other team's work has also been instrumental, as many options are available to classify and quantify drought severity, from indices based on rainfall alone to those that bring in an evaporation-transpiration component [ 11 ]. That research has helped create useful drought indexes employed by monitoring, assessment, and worldwide early warning systems. For example, the Standardised Precipitation Index (SPI) and the SPEI are now important for monitoring global and regional drought and supporting decision-making [ 12 ]. Advancements in analyzing and evaluating drought have suggestively increased, mainly in geospatial modeling, simulation, and prediction for future alterations. Scholars recommend spatially distributed machine learning models, like CA-ANN and ANN-Markov Chain, for analyzing and predicting land changes, land surface temperature (LST), and drought changes. These models offer precise capabilities and exceptional behaviors regarding approach and simulation complexities. Neural network models, especially artificial neural networks (ANNs), are mostly used for variation analysis and prediction because they accurately represent complex geographically heterogeneous land-use and land cover alterations. The coupled CA-ANN model, which combines cellular automata with artificial neural networks, effectively simulates land change systems and utilizes "what-if" situations in land alteration simulation and modeling. In this study, the MOLUSCE plugin, a user-friendly QGIS plugin compatible with versions 2.00 to 2.99, was utilized to estimate future drought conditions and spatial-temporal transitions using SPI and SPEI. This plugin, introduced by Asia Air Survey (AAS) in 2012, offers various algorithms essential for different techniques required in the study's objectives. The study uses MODIS remote sensing, climate data, and a machine learning approach to analyze the spatio-temporal drought disaster over the Ningxia Hui Autonomous Region of China over the last 20 years. The study aims to recognize the temporal-spatial change of drought, assess the effect of drought on agriculture, and determine the characteristics of drought in Ningxia. The study used multi-temporal MODIS imagery to examine spatiotemporal drought deviations in the study area and assess the correctness of the machine learning (ML) approach in predicting drought. The basic objective of this study is to comprehensively analyze drought patterns in the Ningxia Hui Autonomous Region of China from 2003–2023. Thus, the main objectives of this study are 1) Identifying spatiotemporal changes of drought using Multi-Temporal MODIS imagery. 2) Examining the impact of drought on agriculture across the Ningxia. 3) Determining the unique characteristics of drought within the specified region. 4) Implementing machine learning approaches to predict drought existence in the future. This study can provide valuable insights into drought dynamics that may inform improved drought management and agricultural planning approaches. 2 Materials and Methods The RS and machine learning techniques (ML) were applied to predict drought deviations in Ningxia. The process utilizes MODIS & Climate Data (2003–2023) and Precipitation & Evapotranspiration data. Drought and vegetation indices (i.e., KBDI, SPI, SPEI, EVI, and LST) evaluated the variations in drought patterns in the study area. The ANN model was executed to forecast future droughts using SPI and SPEI. The ML component employs a CA-ANN (Convolutional Artificial Neural Network) model, which takes input from the indices SPI (Standardized Precipitation Index) and SPEI (Standardized Precipitation Evapotranspiration Index). The CA-ANN model is used for the future prediction of SPI & SPEI (Drought), ultimately leading to the final map layout as the output of the entire process. The methodology (Fig. 1 ) combines remote sensing data, climate data, image processing techniques, and machine learning algorithms to predict and map drought conditions in a specific study area. 2.1 Study Area Ningxia is an inland province in northwest China, with a land area of about 66,400 square kilometers and over 6 million people (see Fig. 2 ). It is characterized by a dry climate, for which the features of scarce rainfall, high evaporation, and significant differences between day and night temperatures are typical. The annual rainfall lies between 200 to 700 mm and decreases gradually from southeast to northwest. The average yearly precipitation is about 300mm, 60–80% recorded between July and September when East Asian monsoons prevail. The temperature in Ningxia ranges from − 10°C to 25°C, generally revealing the typical continental monsoon climate of the North Temperate Zone [ 13 ]. Its capital is Yinchuan, and its terrain consists mainly of desert, grassland, plains, and mountains, the western Helan Mountains [ 14 ]. The Yellow River flows through the Province. It is a significant water source for irrigation and hydropower and a cause of problems, including flooding, soil erosion, and desertification. 2.2 Datasets A range of satellites and their sensors were used to capture data for 20 years, as shown in Table 1 . The data was taken in 2003, 2022, 2023. The data sources are Earth Explorer, Earth Engine, Climate Data, and DIVA-GIS. The MOD09GA, MOD11A2, and MCD43A4 products are also derived from MODIS data. These products are generated weekly and have a spatial resolution of 500m-1200 km. The study used climate data from Climate Data to capture long-term trend climate conditions affecting their study area. Climate Data is a networked ecosystem for Climate Data. Climate data is the first step in realizing an overall digital earth. Boundary data DIVA-GIS used to determine the study area focus on the area of interest [ 15 ]. Table 1 lists the Datasets used in this study. Satellite/ Product Sensor Year Spatial Resolution Source MODIS MOD09GA, MOD11A2, MCD43A4 2003–2023 500m, 1200*1200km, 463.313m https://earthengine.google.com/ Climate Data Precipitation 2003–2022 Ground Data https://en.climate-data.org/asia/china-110/ Boundary - 2023 - https://www.diva-gis.org/gdata 2.3 Methods 2.3.1 Drought Indices Table 2 The summary of vegetation and drought indices. Product Data Used Spatial Resolution Temporal Resolution Formula LST MOD11A2 1*1 km 8-day \(\:LST=DN\times\:0.02-273.15\) EVI MCD43A4 463.313m 16-Day \(\:EVI=2.5\text{*}\frac{NIR-Red}{(NIR+C1\text{*}Red-C2\text{*}BLUE+L)}\) SPI CHIRPS precipitation 0.05°×0.05° Monthly \(\:=\frac{X-Xm}{\sigma\:}\) SPEI CSIC/SPEI/2_8 0.5° Monthly \(\:SPEI=W-\frac{c0+\text{c}1\text{W}+\text{c}2{W}^{2}}{1+d1W+d2{W}^{2}+d3{W}^{3}}\) KBDI WTLAB/KBDI/v1 4*4km Daily \(\:\text{K}\text{B}\text{D}\text{I}=\text{Q}+\frac{\left(800-Q\right).\left(0.968.{e}^{0.0486.T\:}-8.30\right).\varDelta\:t}{1+10.88.{e}^{0.0486.T\:}}\:.\:{10}^{-3}\) Summary of the vegetation and drought indices is given in the Table 2 along with the formula of each index. Spatial and temporal resolution of indices are different as mentioned in the table. The Enhanced Vegetation Index (EVI) is one of the most commonly used RS indexes for estimating vegetation cover. It is measured using the reflectance values of near-infrared (NIR), red, and blue light from the Earth's surface [ 16 ]. In this study area, research was conducted on the MODIS/MCD43A4 surface reflectance variables on Google Earth Engine (GEE) using EVI. Its ranges typically remain from − 1.0 to 1.0, and higher values show more vegetation. By combining the blue band into the calculation, EVI is less sensitive to atmospheric effects, and its ability to differentiate variations in vegetation cover is improved. $$\:EVI=2.5\text{*}\frac{NIR-Red}{(NIR+C1\text{*}Red-C2\text{*}BLUE+L)}$$ 1 LST is an essential parameter for observing the surface temperature of the Earth and its variations over time. For LST on GEE, the MODIS/061/MOD11A2 dataset was used in the study area giving an average of 8-day 1x1 km LST products. $$\:LST=DN\times\:0.02-273.15$$ 2 ML algorithms find associations and patterns in data, permitting software to recover its performance over time [ 17 ]. The standardized precipitation index (SPI) is widely used to characterize agricultural drought across different timescales [ 18 ]. SPI was computed in the study area of Ningxia by using CHIRPS precipitation data. The script was accomplished through GEE, and the work was split into SPI (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12) calculations. The first calculation was based on the "common" SPI, computed on an n-month basis. A SPI computed on one month would then be referred to as "SPI-1", on six months, "SPI-6," and so on [ 19 ]. The second SPI was computed based on dates of MODIS captures to illustrate changes in precipitation and thus in how the Earth's precipitation is affected by the changes of climate that the globe is currently undergoing [ 20 ]. SPI classification category is given in Table 3 . $$\:SPI=\:-\left(t-\:\frac{{c}_{0}+{c}_{1}t+{c}_{2}{t}^{2}}{1+{d}_{1}t+{d}_{2}{t}^{2}+{d}_{3}{t}^{3}}\right)$$ 3 Table 3 lists SPI Classification [ 21 ] SPI Category Value Less than − 2 Extremely dry Between − 1.5 & -2 Severely dry Between − 1 & -1.5 Dry Between − 0.5 & -1 Moderately dry Between 0.5 & -0.5 Normal Between 0.5 & 1 Wet Between 1 & 1.5 Moderately wet Between 1.5 & 2 Severely wet More than 2 Extremely wet SPEI is an index based on precipitation and evapotranspiration data, recognized as the Standardised Precipitation Evapotranspiration Index (SPEI) [ 22 ], used for the study area, Ningxia. The CSIC/SPEI dataset gives global SPEI data for the entire Earth at a spatial resolution of 0.5º [ 23 ]. A 1-month, 3-month, 6-month, 9-month, and 24-month monitoring was done in the study area, Ningxia. The dataset monitored drought fluctuations (wet and dry spells) from 2003 to 2023 in the study area. Different categories of SPEI are classified as discussed in Table 4 . $$\:SPEI=W-\frac{{c}_{0}+{c}_{1}\text{W}+{c}_{2}{W}^{2}}{1+{d}_{1}W+{d}_{2}{W}^{2}+{d}_{3}{W}^{3}}$$ 4 Where: \(\:W=\:\sqrt{-2\text{ln}\left(P\right)\:\:}\) for P ≤ 0.5 (5) P = probability of exceeding a determined D value, \(\:p=1-f\left(x\right);\) When P > 0.5, \(\:p=1-P,\) constants are: \(\:{c}_{0}\:=\:2.515517\) \(\:{d}_{1}\:=\:1.432788\) \(\:\:{c}_{1}\:=\:0.802853\) \(\:{d}_{2}\:=\:0.189269\) \(\:\:{c}_{2}=\:0.010328\) \(\:{d}_{3}=\:0.001308\) Table 4. SPEI Classification (Fu et al. 2022). SPEI Category Value Extremely wet More than 2.00 Very wet 1.50 to 1.99 Moderately wet 1.00 to 1.49 Near Normal -0.99 to 0.99 Moderately dry -1.00 to -1.49 Severely dry -1.50 to -1.99 Extremely dry Less than − 2.00 Daily maximum temperature and precipitation measurements determine the Keetch-Byram drought index (KBDI) that bears on evapotranspiration [ 24 ]. The daily maximum temperature enables calculating the amount of water evaporated from the soil surface evaporative demand and total precipitation, from which KBDI, the soil moisture deficit, and the amount needed to bring a site's soil moisture to field capacity can be determined. The UTOKYO/WTLAB/KBDI/v1 dataset, which provides a continuous reference scale that divides the moisture regime of the soil and duff layers into eight classes from 0.0 (no moisture deficit) to 800.0 (extreme drought) by the cumulative drying that occurs during each day of no rain (Table 5 ), the rate of which, in turn, depends on the daily highs [ 25 ]. KBDI values provide a relative measure of soil moisture and fire risk, making them valuable indicators in assessing the impact of drought on potential wildfire hazards. Table 5 Lists KBDI Classification [ 26 ]. KBDI Value Category 0 to 200 Indicates high soil moisture, suggesting a lower risk of wildfire in the presence of ample water content. 200 to 400 It represents moderate soil moisture, signifying a moderate wildfire risk, especially in regions experiencing drought. 400 to 600 Reflects low soil moisture, indicating an elevated wildfire risk, particularly in drought areas. 600 to 800 It signifies deficient soil moisture, highlighting an extreme wildfire risk, particularly in regions undergoing severe drought. $$\:\text{K}\text{B}\text{D}\text{I}\:=\text{Q}+\frac{\left(800-Q\right).\left(0.968.{e}^{0.0486.T\:}-8.30\right).\varDelta\:t}{1+10.88.{e}^{0.0486.T\:}}\:.\:{10}^{-3}$$ 6 Q, which represents the previous day's KBDI adjusted by the net rainfall in inches per hundred (cf. details below); T, the air temperature in degrees Fahrenheit; Δt, the time increase (typically one day); and P, signifying the mean annual precipitation in inches. $$\:Q={KBDI}_{t-1\:}-\:\text{P}{net}_{t}\:.\:100$$ 7 $$\:\text{P}{net}_{t}=\text{m}\text{a}\text{x}[0,{P}_{t\:}-\text{m}\text{a}\text{x}(0,{P}_{\text{l}\text{i}\text{m}\:}-{\sum\:}_{i=1}^{rr-1}{P}_{\text{t}-\text{i}\:})$$ 8 With "rr" denoting the count of consecutive days on which rain has occurred. The agricultural drought disaster model summary using the machine learning model (MLM) is in Table 6 . Various MLMs are used for drought prediction, depending on the available data. For example, meteorological data and a combination of CART and SVM models are used to predict SPEI annually. In contrast, CART Cubist models can use MODIS data to predict SPI on a seasonal basis, including early, growing, middle, and late seasons. LSTM models can predict SPEI for up to 12 months based on soil moisture, LST, ET, EVI, and precipitation [ 27 ]. Table 6 lists the Agricultural drought prediction summary using machine learning models (MLMs). Model Data type Forecaster variables Response variable Predicting lead time Outcome CART and SVM MODIS EVI, NDVI, LST SPEI Seasonal Increased drought area prediction SVM Meteorological data Slope, aspect, elevation, annual precipitation SPI - Agricultural drought prediction Cubist MODIS, TRMM and climate data EVI, LST SPI Seasonal Severe Drought Index Mapping SVM Soil Moisture LST, ET, EVI, precipitation SPEI and crop yield 12-months Drought severity distribution maps 2.3.2 Change Evaluation and Future Prediction using SPI and SPEI In this study, the MOLUSCE plugin in QGIS was employed to simulate the SPI and SPEI change between their classes (extremely dry, severely dry, moderately dry, normal, and wet) and estimate spatiotemporal changes of drought for the periods 2003 to 2023. SPI (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12) and SPEI (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12) maps of each year were produced. The 2003 and 2023 drought (SPI and SPEI) created area variation and transition probability matrixes. The artificial neural network (ANN) multilayer perception strategy was implemented. Precipitation, temperature, and PET were taken as the determinant factors for future drought variation prediction. In drought change analysis and prediction, these variables are frequently used because they deliver verifiable information on the effect of anthropogenic and natural factors on SPI and SPEI variations. 3 Results 3.1 Enhanced Vegetation Index (EVI) changes from 2003–2023 The maximum EVI in 2003 did not exceed 0.38, whereas the maximum in 2023 was 0.33, as shown in Fig. 3 . The maps in Fig. 3 show that EVI in Ningxia predominantly decreased over the past 20 years. Climate change has caused a significant incline in the frequency and severity of drought. This puts the region at greater risk of soil erosion. 2003 provides an early baseline for pre-drought assessment, while 2018 and 2023 show more recent conditions after prolonged drought. The selected years deliver good temporal sampling aligned with significant droughts. These are all valuable factors for generating a robust EVI trend analysis for assessing desertification. Machine learning has been used to study desertification change in Ningxia. From 2003 to 2023, the average annual soil erosion rate in Ningxia, China, increased [ 28 ]. 3.2 Land Surface Temperature (LST) changes from 2003–2023 The map below shows the Land Surface Temperature (LST) of Ningxia, China, from 2003 to 2023. The LST is in degrees Celsius (°C). The LST is 4.6°C to 31°C, as displayed in Fig. 4 . The LST of Ningxia has been increasing with time. In 2003, the average LST of Ningxia was 14.3°C. The average LST of Ningxia will be 16.7°C by 2023. The inclination of the trend line in Fig. 5 shows the increase in the temperature from 31°C to 34°C from 2003 to 2023. The years were likely chosen to provide evenly spaced data points over the 20 years to see the overall trend in LST over time. 2003 provides an early baseline temperature, while 2006, 2012, 2015, 2018, and 2023 allow for comparison of the LST approximately every 3, 6, 9, and 15 years after the initial 2003 data point. 3.3 Keetch-Byram Drought Index (KBDI) changes from 2003–2023 KBDI is a measure used to assess the dryness of an area and the potential risk of wildfire. In Fig. 6 , the spatial distribution of KBDI in Ningxia changes over time. In 2007, the KBDI of Ningxia was 580; by 2023, the average KBDI of Ningxia had increased to 680. This increase in KBDI indicates that Ningxia has become drier over time and that the risk of wildfires has increased. Figure 7 shows that drought tends to grow from 2007 to 2023. 3.4 Standardized Precipitation Index (SPI) changes from 2003–2023 Ningxia experienced various wet and dry conditions over the 2003 to 2023 period, as illustrated in Fig. 8 . Most of the years experienced drought in the north part of the region in 2003, 2005, 2013, 2014, 2017, 2019, and 2023. However, the opposite pattern occurred in 2004, 2007, 2009, 2011, 2012, 2016, and 2022, which shows the extreme drought experienced in the southern part of Ningxia. SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12 variations of the drought pattern in Ningxia from 2003–2023 are displayed in Fig. 9 . The timescales of SPI determine the wet changes, and the negative (decrease) in the timescales demonstrates the dry spells. In SPI-1 2003, the dry spell ranges from 0.7 to -1, and the wet spell is from 1 to 1.3. 2014, 2018, 2021 and 2022. SPI-3 has a high damp spell in 2013 and 2018, while dry spells with − 1.5 to -2 are in 2003, 2011,2021, and 2023. SPI-6 has more dry spell fluctuations, as in 2006, 2007, and 2022, and SPI is between − 2 to -2.4. The wet spell of SPI-9 reached + 2 to + 3 in 2015 and 2018, while the highest dry spell was − 1.5 in 2021. SPI-12 demonstrates more dry spells in 2005, 2006, 2011, 2015, 2019, 2021, and 2022. 3.5 Standardized Precipitation Evapotranspiration Index (SPEI) In 2003, most of the region experienced normal conditions. In 2004, 2005, 2015, 2018, and 2019, the SPEI map shows that the northeastern half of Ningxia experienced extremely dry conditions, while in 2006, the northern half of the area experienced extreme conditions. Overall, more drought years were 2004, 2005, 2006, 2009, 2015, 2016, 2018, 2019, and 2021. SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12 changes in the drought pattern in Ningxia from 2003–2023 are displayed in Fig. 11 . Different timescales of SPEI determine the variations in the wet and dry spells in Ningxia. In SPEI-1, the damp spell ranges from 1 to 2 in 2007, 2008, 2016, and 2018, while the dry spell is between − 0.7 to -2.8 in 2004, 2009, 2015, and 2021. SPEI-3 had the highest wet spell in 2012 and the highest dry spell in 2015. SPEI-6 has frequent dry spells in 2004, 2005, 2006, 2009, 2015, 2020, and 2023. The 9-month scale (SPEI-9) reflects the long-term drought process with more extreme droughts in 2003, 2005, 2006, 2009, 2011, 2013, and 2015. The 12-month scale (SPEI-12) of SPEI reflects more frequent extreme droughts in 2005, 2006, 2009,2011, 2015, 2020, 2022, and 2023. 3.6 Drought Change and Prediction Using the CA-ANN Model The future prediction, along with the previous (2003, 2013) and recent (2023) map of SPI in Fig. 12 , shows predicted dryness or drought conditions for Ningxia in the year 2033 in (a4, b4, c4, d4, and e4). The map uses the Standardised Precipitation Index (SPI) to predict drought conditions. The map of SPI (SPI-1, SPI-3, SPI-6, SPI-9, SPI-12) indicates that the areas at risk of drought in 2033 are northwest Ningxia, the central Plains, and the northeast. SPI-12 shows more frequent severe and extreme drought in the upper north and east part of the region for 2033. The future prediction result in Fig. 13 of SPEI indicates predicted drought conditions for 2033 in (a4, b4, c4, d4, e4). The map uses the SPEI (SPEI-1, SPEI-3, SPEI-6, SPEI-9, SPEI-12) to predict drought conditions. According to future predictions, the areas at risk of drought in (a) SPEI-1 are the north and western Ningxia in (a1) and (a4). SPEI-1 of 2013 in (a2) had an extreme drought in the north region, while the southern region of Ningxia is moderately dry. The (b) SPEI-3 and (c) SPEI-6 display the northern and southern regions of Ningxia to be extremely dry in (b4), and in (c4), extreme drought is predicted to be in the western region of Ningxia. SPEI-9 (d4) shows Ningxia's northeastern and southwestern regions to be severely dry. In SPEI-12, the (e4) is expected to be extremely dry in the western and southern parts in 2033. The kappa coefficient for SPI and SPEI timescale predicted results employing the CA-ANN Model is given in Table 7 . The results show that the SPEI-6 had the highest degree of agreement (96%), and the SPEI-9 had the lowest degree of agreement (65%). Table 7 Future Prediction for SPI and SPEI. Current Predicted Kappa Value for 2033 2023 2033 CA-ANN Model Validation SPI Timescale SPI-1 -1.835 -2.00 0.82965 0.93093 SPI-3 -0.713 -1.355 0.81178 0.93639 SPI-6 -1.613 -2.475 0.93243 0.95226 SPI-9 -0.555 -1.578 0.86640 0.95697 SPI-12 -1.320 -1.991 0.91596 0.95360 SPEI Timescale SPEI-1 -1.898 -2.394 0.93085 0.86665 SPEI-3 -1.127 -1.365 0.88649 0.81905 SPEI-6 -1.622 -1.827 0.83258 0.96389 SPEI-9 -1.369 -1.735 0.73214 0.65756 SPEI-12 -2.046 -2.894 0.92242 0.88763 4 Discussion Properly monitoring drought has significant implications for mapping drought spatial anomalies over time, risk management, drought preparedness, and agricultural irrigation for the future [ 29 ]. This study also assessed the changes in drought patterns yearly from 2003 to 2023. In this research, remote sensing and machine learning approaches are used to get effective drought monitoring results, which devastates agriculture and threatens the food security of China's Ningxia Hui Autonomous Region. Remotely sensed vegetation and drought monitoring combined with machine learning to understand unprecedented drought intensity patterns across Ningxia during the study period. An Enhanced Vegetation Index (EVI) was performed to monitor vegetation anomalies over time. Land Surface Temperature (LST) is estimated to analyze the correlation of temperature with drought. Standardized Precipitation Index (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12), Standardized Precipitation Evapotranspiration Index (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12), and Keetch-Byram Drought Index (KBDI) were used as drought indicators for drought monitoring across study area (Fig. 2 ). The results of EVI quantify the impacts of vegetation on health, which showed that 2003 experienced the highest vegetation, ranging from 0.27 to 0.38. In contrast, the range of vegetation in 2023 was 0.24 to 0.33, as shown in Fig. 3 . The results of EVI claim that vegetation has declined in Ningxia over the previous two decades. The assessment of LST indicated that in 2003, the highest temperature was about 31°C, which increased to 38°C in 2023. The increase in temperature was about 4°C in the two decades. Fi 4 displays the increase in land surface temperature of Ningxia from 2003 to 2023. The rise in temperature is shown in Fig. 5 , where the trend line determines the inclination in the temperature. EVI and LST show an inverse relation with each other. Our findings suggest that the increased LST can be due to decreased vegetation in the Ningxia. [ 30 ] also investigated how increased temperature reduces vegetation and vice versa. The index KBDI indicates drought intensities from soil moisture deficits [ 31 ]. Multi-sensor and multi-index spatial analyses delineate drought risks spatially and intensity-wise heterogeneous nature across Ningxia. At the same time, most of Ningxia reveals moderate to severe drought conditions, the highest drought intensity over most indicators clustered centrally. KBDI flags this region as having the most severe and agriculturally impactful drought severity, as shown in Fig. 6 and Fig. 7 ; soils dry out from water scarcity in the area historically. Furthermore, the results of SPI illustrated that the overall SPI noticeably decreased from 2003 to 2023. The SPI indicated that the most affected years were 2005, 2006, 2009, 2012, 2016, 2020, and 2022, as shown in Fig. 8 . SPEI indicated the study area has become drier in several years, and Fig. 10 shows years of severe and extreme droughts in the years 2004, 2005, 2006, 2009, 2015, 2016, 2018, 2019, and 2021. SPI and SPEI quantify drought from precipitation deficits [ 21 ]. Our study found the most robust relationship between SPEI and KBDI, suggesting that in addition to precipitation deficits by SPEI, capturing the evapotranspiration dynamics gave a critical indication that dry enough soil moisture affects agriculture. This relationship was also found in the relevant study by (Thanh ) in 2023. EVI has weaker relationships, indicating the vegetation response to drought depends on the specific vegetation species, farming practices, and environmental micro-climates [ 32 ]. Moreover, the study suggested that SPEI was the most efficient indicator and gave better results than SPI and KBDI. This indicator is also considered superior in the study conducted by [ 33 ]. The machine learning algorithm integrated multi-index datasets. CA-ANN model provides a more detailed and robust agricultural drought intensity zoning across the spatiotemporal scales of Ningxia. These algorithms are used because of their simplicity, effectiveness, and reliability. The algorithm identified central Ningxia as a zone of recurring high drought severity, again showing the strength of the spatial monitoring that comes from integrating multiple indices. In addition to that, Fig. 12 shows the comparison of the previous and recent decades with the future prediction of drought conditions for the year 2033 using SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12. It shows that northwest and northeastern Ningxia are at risk of facing extreme drought conditions. SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12 for 2033 also show that some areas of Ningxia are at risk of drought. These areas are north, west, south, northeastern, southwestern, and northwestern parts of Ningxia, as shown in Fig. 13 . Thus, remote sensing and machine learning can be used for drought monitoring and assessing its impacts on the relevant [ 34 ]. In future studies, testing group performance can be evaluated by integrating the Root Mean Square Error (RMSE) technique. Future predictions of drought can also help in drought preparedness and recovery. 5 Conclusion This work evaluated the temporal-spatial variation of drought in Ningxia Hui Autonomous Region, China. The results show an increase in severe and moderate drought areas of 15% and 10%, while there was a 25% decrease in mild drought areas in the 2003–2023. These results can guide the region's direction regarding drought mitigation and adaptation strategies. The study area is a critical food production area in northwest China that is severely affected by drought. MODIS reflectance data, LST, precipitation, evapotranspiration data, and meteorological data from 14 meteorological stations were used to assess the temporal-spatial variation of drought. The results showed that applying drought indices and using machine learning, like the CA-ANN model, represents a comprehensive utilization approach for discerning and categorizing drought severity levels. This study is significant given the acceleration of severe and moderate drought areas and the decline of mild drought areas in the study area during this two-decade timeframe. This spatial and time-based reasoning is necessary to inform the region's policy and decision-makers responsible for drought management and agricultural activities. This approach and its findings add to the required directions to monitor and mitigate the influence of drought on the environment and economy of the Ningxia Hui Autonomous Region. Declarations Author Contribution M.A.K., data curation, investigation, software, code, writing—original draft; J.Z., conceptualization, funding acquisition, supervision; writing—review; S.A., software, visualization, writing—review; Z.Z., code, visualization, writing—review; H.U., visualization, writing—review; A.G., writing—review. All authors have read and agreed to the published version of the manuscript Acknowledgements. This work was supported by the Central Guiding Local Science and Technology Development Fund of Shandong-Yellow River Basin Collaborative Science and Technology Innovation Special Project (No. YDZX2023019). References G. Zhang, X. Su, O. O. Ayantobo, and K. Feng, “Drought monitoring and evaluation using ESA CCI and GLDAS-Noah soil moisture datasets across China,” Theor Appl Climatol, vol. 144, no. 3, pp. 1407–1418, 2021, doi: 10.1007/s00704-021-03609-w. Q. Zhang et al., “Causes and Changes of Drought in China: Research Progress and Prospects,” Journal of Meteorological Research, vol. 34, no. 3, pp. 460–481, 2020, doi: 10.1007/s13351-020-9829-8. Y. Xu et al., “Drought assessment of China in 2002–2017 based on a comprehensive drought index,” Agric For Meteorol, vol. 319, p. 108922, 2022, doi: https://doi.org/10.1016/j.agrformet.2022.108922. P. Sun, Z. Ma, Q. Zhang, V. P. Singh, and C.-Y. Xu, “Modified drought severity index: Model improvement and its application in drought monitoring in China,” J Hydrol (Amst), vol. 612, p. 128097, 2022, doi: https://doi.org/10.1016/j.jhydrol.2022.128097. W. Liu, F. Sun, S. Sun, L. Guo, H. Wang, and H. Cui, “Multi-scale assessment of eco-hydrological resilience to drought in China over the last three decades,” Science of The Total Environment, vol. 672, pp. 201–211, 2019, doi: https://doi.org/10.1016/j.scitotenv.2019.03.408. Q. Zhang et al., “Droughts across China: Drought factors, prediction and impacts,” Science of The Total Environment, vol. 803, p. 150018, 2022, doi: https://doi.org/10.1016/j.scitotenv.2021.150018. Y. Wang, W. Zhao, Q. Zhang, and Y. Yao, “Characteristics of drought vulnerability for maize in the eastern part of Northwest China,” Sci Rep, vol. 9, no. 1, p. 964, 2019, doi: 10.1038/s41598-018-37362-4. L. I. U. Bin, P. Li, Z. Liping, and R. E. N. Shuyong, “Drought Resistance Study of 10 Major Ornamental Shrub in Ningxia,” Acta Botanica Boreali-Occidentalia Sinica, vol. 33, no. 9, pp. 1808–1816, [Online]. Available: xbzwxben/article/abstract/20130914 D. Xu, Y. Ding, H. Liu, Q. Zhang, and D. Zhang, “Applicability of a CEEMD–ARIMA Combined Model for Drought Forecasting: A Case Study in the Ningxia Hui Autonomous Region,” Atmosphere (Basel), vol. 13, no. 7, 2022, doi: 10.3390/atmos13071109. C. Wang et al., “Impacts of drought on maize and soybean production in northeast China during the past five decades,” Int J Environ Res Public Health, vol. 17, no. 7, Apr. 2020, doi: 10.3390/ijerph17072459. C. Tan, J. Yang, X. Wang, D. Qin, B. Huang, and H. Chen, “Drought disaster risks under CMIP5 RCP scenarios in Ningxia Hui Autonomous Region, China,” Natural Hazards, vol. 100, no. 3, pp. 909–931, 2020, doi: 10.1007/s11069-019-03811-0. T. Wu, B. Li, L. Lian, Y. Zhu, and Y. Chen, “Assessment of the Combined Risk of Drought and High-Temperature Heat Wave Events in the North China Plain during Summer,” Remote Sens (Basel), vol. 14, no. 18, Sep. 2022, doi: 10.3390/rs14184588. Z. Liu et al., “An Analysis of the Spatiotemporal Variation in Wind Erosion–Climate Erosion Force in the Ningxia Region and Its Driving Factors,” Atmosphere (Basel), vol. 14, no. 6, Jun. 2023, doi: 10.3390/atmos14060963. Y. ZHENG, H. MENG, X. XIE, and S. SHI, “Dynamics of Climate Change, Drought and Migration: A Mixed Method Research in Ningxia,” Chinese Journal of Urban and Environmental Studies, vol. 05, no. 03, p. 1750021, Sep. 2017, doi: 10.1142/s234574811750021x. X. Zhao, H. Xia, B. Liu, and W. Jiao, “Spatiotemporal Comparison of Drought in Shaanxi–Gansu–Ningxia from 2003 to 2020 Using Various Drought Indices in Google Earth Engine,” Remote Sens (Basel), vol. 14, no. 7, 2022, doi: 10.3390/rs14071570. R. Singh, M. Kumari, S. Bindal, and I. Gupta, “Quantification of Drought Condition Using Drought Indices: A Review,” in Climate Change and Water Security, S. Kolathayar, A. Mondal, and S. C. Chian, Eds., Singapore: Springer Singapore, 2022, pp. 231–241. Y. Zhao et al., “Drought Monitoring and Performance Evaluation Based on Machine Learning Fusion of Multi-Source Remote Sensing Drought Factors,” Remote Sens (Basel), vol. 14, no. 24, Dec. 2022, doi: 10.3390/rs14246398. Z. Pei, S. Fang, L. Wang, and W. Yang, “Comparative analysis of drought indicated by the SPI and SPEI at various timescales in inner Mongolia, China,” Water (Switzerland), vol. 12, no. 7, Jul. 2020, doi: 10.3390/w12071925. Q. Zhang, J. Li, V. P. Singh, and Y. Bai, “SPI-based evaluation of drought events in Xinjiang, China,” Natural Hazards, vol. 64, no. 1, pp. 481–492, 2012, doi: 10.1007/s11069-012-0251-0. Q. He, M. Wang, K. Liu, B. Li, and Z. Jiang, “Spatiotemporal analysis of meteorological drought across China based on the high-spatial-resolution multiscale SPI generated by machine learning,” Weather Clim Extrem, vol. 40, p. 100567, 2023, doi: https://doi.org/10.1016/j.wace.2023.100567. C. Tan, J. Yang, and M. Li, “Temporal-spatial variation of drought indicated by SPI and SPEI in Ningxia Hui Autonomous Region, China,” Atmosphere (Basel), vol. 6, no. 10, pp. 1399–1421, Oct. 2015, doi: 10.3390/atmos6101399. H. Zhao, G. Gao, W. An, X. Zou, H. Li, and M. Hou, “Timescale differences between SC-PDSI and SPEI for drought monitoring in China,” Physics and Chemistry of the Earth, vol. 102, pp. 48–58, Dec. 2017, doi: 10.1016/j.pce.2015.10.022. C. Hao, J. Zhang, and F. Yao, “Combination of multi-sensor remote sensing data for drought monitoring over Southwest China,” International Journal of Applied Earth Observation and Geoinformation, vol. 35, no. PB, pp. 270–283, 2015, doi: 10.1016/j.jag.2014.09.011. W. Takeuchi et al., “Indonesia 4 Institute of Meteorology, Hydrology and Environment (IMHEN), Vietnam 5 Myanmar Geospatial and Resource Program (MGRP),” MRC, 2017. [Online]. Available: http://sharaku.eorc.jaxa.jp/GSMaP M. Häusler, J. P. Nunes, J. M. N. Silva, J. J. Keizer, T. Warneke, and J. M. C. Pereira, “A promising new approach to estimate drought indices for fire danger assessment using remotely sensed data,” Agric For Meteorol, vol. 274, pp. 195–209, Aug. 2019, doi: 10.1016/j.agrformet.2019.04.015. K. Shojaeizadeh, M. Ahmadi, and A. Dadashi-Roudbari, “Investigating drought in Iran using Keetch-Byram drought index (KBDI) and evaluating it with surface soil moisture (SSM) of active microwave sensors,” Iranian Journal of Geophysics, vol. 17, no. 2, p. 240, Jul. 2023, doi: 10.30499/IJG.2023.391095.1509. G. Ghazaryan, O. Dubovyk, V. Graw, N. Kussul, and J. Schellberg, “Local-scale agricultural drought monitoring with satellite-based multi-sensor time-series,” GIsci Remote Sens, vol. 57, no. 5, pp. 704–718, Jul. 2020, doi: 10.1080/15481603.2020.1778332. B. Su et al., “Drought losses in China might double between the 1.5 °C and 2.0 °C warming,” Proceedings of the National Academy of Sciences, vol. 115, no. 42, pp. 10600–10605, Oct. 2018, doi: 10.1073/pnas.1802129115. J. Qiu, Z. Shen, and H. Xie, “Drought impacts on hydrology and water quality under climate change,” Science of The Total Environment, vol. 858, p. 159854, 2023, doi: https://doi.org/10.1016/j.scitotenv.2022.159854. E. Li, J. Zhao, W. Zhang, and X. Yang, “Spatial-temporal patterns of high-temperature and drought during the maize growing season under current and future climate changes in northeast China,” J Sci Food Agric, vol. 103, no. 12, pp. 5709–5716, Sep. 2023, doi: 10.1002/jsfa.12650. F. Zhao and Y. Liu, “Important meteorological predictors for long-range wildfires in China,” For Ecol Manage, vol. 499, p. 119638, 2021, doi: https://doi.org/10.1016/j.foreco.2021.119638. H. Jia, F. Chen, E. Du, and L. Wang, “Drought Vulnerability Curves Based on Remote Sensing and Historical Disaster Dataset,” Remote Sens (Basel), vol. 15, no. 3, 2023, doi: 10.3390/rs15030858. C. A. Alonzo, J. M. Galabay, M. N. Macatangay, M. B. Magpayo, and R. Ramirez, “Drought Risk Assessment and Monitoring of Ilocos Norte Province in the Philippines Using Satellite Remote Sensing and Meteorological Data,” AgriEngineering, vol. 5, no. 2, pp. 720–739, 2023, doi: 10.3390/agriengineering5020045. J. Li et al., “Robust Meteorological Drought Prediction Using Antecedent SST Fluctuations and Machine Learning,” Water Resour Res, vol. 57, no. 8, Aug. 2021, doi: 10.1029/2020WR029413. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-5259358","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":366002026,"identity":"e3f6c36b-4016-4888-aab1-10c254b95bf9","order_by":0,"name":"Muhammad Awais Khan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBklEQVRIie3PsUrEMBjA8e8oeEvANQ72nkBIKfSW83yVfhTq0sHxBoeDgl36AB3EZwgIBSdTArklD1CoYLs4OdSto1FcHCJ1uyH/JQTy48sH4HIdaQKEAFjmQsQ7c10CLPazCFHY99qc3gwC34RmYTDczSCnVTI0k35ZrYlKKT5s8NnzmqGCrX9hcbRNmSTtW/BU5IpinWLpnSQBhySMhGVMFzMJo1xw/TWlloaQ6KwHgbWFrLrrsZlGecXbLKJ4P4OwLmOCtBINCRnufwj/gwSv7zeSaJmYj2EfqzQspdmlYvZdfH14/JiUvOSHXDTT7ea8KPJmKHdb37q+Jfa/5y6Xy+X63SctCnIRQGVMEgAAAABJRU5ErkJggg==","orcid":"","institution":"Qingdao Univeristy","correspondingAuthor":true,"prefix":"","firstName":"Muhammad","middleName":"Awais","lastName":"Khan","suffix":""},{"id":366002028,"identity":"cd6c7da7-c1ce-4c0f-b768-e7ef09fa8ba4","order_by":1,"name":"Shawkat Ali","email":"","orcid":"","institution":"Qingdao Univeristy","correspondingAuthor":false,"prefix":"","firstName":"Shawkat","middleName":"","lastName":"Ali","suffix":""},{"id":366002030,"identity":"2cdb62af-0189-4fc1-9c30-756dccf923da","order_by":2,"name":"Zakria Zaheen","email":"","orcid":"","institution":"Qingdao Univeristy","correspondingAuthor":false,"prefix":"","firstName":"Zakria","middleName":"","lastName":"Zaheen","suffix":""},{"id":366002031,"identity":"90324c7c-be30-413d-8a6c-5c4b1a576d87","order_by":3,"name":"Hidayat Ullah","email":"","orcid":"","institution":"Qingdao Univeristy","correspondingAuthor":false,"prefix":"","firstName":"Hidayat","middleName":"","lastName":"Ullah","suffix":""},{"id":366002032,"identity":"2f5d4ae5-461b-4e31-b27e-bbc382adbf5e","order_by":4,"name":"Amina Gul","email":"","orcid":"","institution":"Qurtab University of Science and Information Technology","correspondingAuthor":false,"prefix":"","firstName":"Amina","middleName":"","lastName":"Gul","suffix":""},{"id":366002035,"identity":"aa9b40ca-e9dd-4b73-b47c-a78f2b29c736","order_by":5,"name":"Jiahua Zhang","email":"","orcid":"","institution":"Qingdao Univeristy","correspondingAuthor":false,"prefix":"","firstName":"Jiahua","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2024-10-14 08:23:44","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5259358/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5259358/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":66805888,"identity":"dac1ff29-2666-4645-ad70-fe3a7bc65ce4","added_by":"auto","created_at":"2024-10-16 15:50:53","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":115366,"visible":true,"origin":"","legend":"\u003cp\u003eMethodology Flowchart\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/06e91f5102f951a310dea559.png"},{"id":66805701,"identity":"b0b704ea-9d4c-4006-9487-d3df005855ee","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":314966,"visible":true,"origin":"","legend":"\u003cp\u003eNingxia Hui Autonomous Region, China. (The core map focuses on the digital elevation model status of Ningxia, while the secondary map shows the exact location of Ningxia in China).\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/03c78b6f831064273473ef40.png"},{"id":66806727,"identity":"d380bd90-dc95-44f4-ae48-7ee53f224c5e","added_by":"auto","created_at":"2024-10-16 15:58:53","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1399628,"visible":true,"origin":"","legend":"\u003cp\u003eEVI Map from 2003-2023\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/303ca60cd45e2c20955d5a4a.png"},{"id":66805707,"identity":"d74eb0b9-9858-4170-9f44-ee22b30530d4","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1296147,"visible":true,"origin":"","legend":"\u003cp\u003eLST Map from 2003-2023\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/91c319fe6770d10bbbc98dbd.png"},{"id":66805704,"identity":"f9fed7c5-e8c4-42a5-a745-60209add1555","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":488811,"visible":true,"origin":"","legend":"\u003cp\u003eLST graph displaying the incline in temperature of land surface from 2003 to 2023\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/93feba9673c10a3da9493a11.png"},{"id":66805889,"identity":"7569a322-1e66-4a76-bd6f-c57577091387","added_by":"auto","created_at":"2024-10-16 15:50:53","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1129082,"visible":true,"origin":"","legend":"\u003cp\u003eKBDI Maps from 2007-2023\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/eedcd39892d4f52f55531b61.png"},{"id":66805703,"identity":"87657edc-1835-4a6d-ab8e-b59f00525eea","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":185933,"visible":true,"origin":"","legend":"\u003cp\u003eThe Keetch-Byram Drought Index (KBDI) graph from 2007-2023\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/2a8ca980ecb8a5811897fafe.png"},{"id":66806728,"identity":"895adfc5-70b2-49ac-8987-c61b71e260cb","added_by":"auto","created_at":"2024-10-16 15:58:53","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":1054708,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of SPI-12 from 2003 to 2023 over Ningxia\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/8c60b2bfd8c2020779d6d65e.png"},{"id":66805891,"identity":"ede27ccf-4809-45eb-9918-a377dc384be2","added_by":"auto","created_at":"2024-10-16 15:50:53","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":333103,"visible":true,"origin":"","legend":"\u003cp\u003eSPI timescales distribution trends from 2003-2023 in Ningxia\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/e06778f820ea67a15cbb1e85.png"},{"id":66807358,"identity":"97355008-b328-44c3-92bc-1c726c5d5ca8","added_by":"auto","created_at":"2024-10-16 16:06:53","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":201008,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial distribution of SPEI-12 from 2003-2023\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/2e52516a6f4f7410b37b4aaf.png"},{"id":66805893,"identity":"23776c23-6cd4-4b17-9e46-985853ba3fa8","added_by":"auto","created_at":"2024-10-16 15:50:53","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":338722,"visible":true,"origin":"","legend":"\u003cp\u003eSPEI variations from 2003-2023\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/14f67496401b7fb6cfdfaa82.png"},{"id":66805712,"identity":"42fa3ec9-45fa-4447-ba2a-368ff9c9166f","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":827478,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of predicted (a) SPI-1, (b) SPI-3, (c) SPI-6, (d) SPI-9, and (e) SPI-12\u003c/p\u003e","description":"","filename":"12.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/93411a423bee23b3e4cde210.png"},{"id":66805711,"identity":"32672cb2-c225-4814-bbfd-5ca6984385dc","added_by":"auto","created_at":"2024-10-16 15:42:53","extension":"png","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":797124,"visible":true,"origin":"","legend":"\u003cp\u003eComparison of predicted SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12\u003c/p\u003e","description":"","filename":"13.png","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/003e1bc3526e35eac0d4bc39.png"},{"id":68055334,"identity":"f41db8f5-07cc-4f99-baf8-497a3b45ea65","added_by":"auto","created_at":"2024-11-02 01:01:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10298762,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5259358/v1/35d4293e-64f9-409c-ba1d-95d0cb9644a2.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Spatio-temporal Monitoring of Drought using Machine Learning approach and Remote Sensing Techniques in Ningxia","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eDrought is an intricate and recurring natural disaster that influences agricultural production, water resources, ecology, and socio-economic development across scales from local communities to regions. Severe droughts can decimate crop yields, grazing lands, and livestock herds, stress water supplies, fuel wildfires, and lead to large economic losses [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Many regions of the world face more pronounced and, in some cases, more recurrent droughts as climate change changes historical drought patterns and their intensity. Understanding the historical variations of drought and its susceptibilities is imperative to facilitate adaptation and mitigation approaches [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Ningxia is in arid northwest China and has been prone to severe drought due to its desert climate with high temperatures and low annual precipitation. Foremost droughts have led to drinking and stock water deficiencies in recent decades and declined agricultural output in Ningxia and the surrounding areas [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. In this study, we perform an inclusive spatio-temporal analysis of drought patterns in Ningxia from the last two decades (2003\u0026ndash;2023). We combine remote sensing (MODIS) data and climate records of precipitation, temperature, evapotranspiration, and soil moisture to utilize MODIS constant exposure of vegetation health and drought stress across Ningxia from 2003 to 2023 and the temporal patterns in drought-related variables across the study area over the same period. Then, we use machine learning methods, such as random forests (RF) and regression trees (RTs), to model the correlations between the MODIS inputs and the severity of the drought throughout Ningxia [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Drought calamities, characterized by prolonged scarcities in precipitation, represent complex natural phenomena that considerably affect the environment, agriculture, and human society [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. They are closely connected to the continuously changing climate system, and their occurrence and severity are deeply affected by variations in the atmosphere and persistent weather patterns. The combination of increasing temperatures and an ongoing decline in precipitation from long periods initiates a gradual reduction in soil moisture, lessening water supplies, and a loss of agricultural productivity. The vulnerabilities of regions already prone to dry conditions or relying on a scarce inventory of water resources are magnified with the arrival of drought, which can destroy food supplies, livelihoods, and freshwater availability. The modern period of climate change especially leads to a greater reappearance and severity of drought events across many regions of the world, with droughts becoming more recurrent and severe [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Understanding the complex interchange of drought and climate is of great scientific interest and is serious for effective disaster management and the creation of adaptable approaches. Such an understanding delivers the foundation for mitigating the multiplicity of effects connected with droughts, ensuring supportable agricultural techniques, and creating the resilience and health of both communities and ecosystems, as they face the enhancing challenge of these disastrous climate procedures [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Writing at the time of that original column, another team strained the need to nail down the definitions of drought, pointing out its individual and multidimensional nature [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. They called for a holistic method that considered meteorological, hydrological, and socio-economic components to improve understanding of drought [\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. This multispectral perspective has become one of the fundamentals of drought research, delivering more detailed evaluations and the ability to more effectively craft mitigation plans [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Meanwhile, the other team's work has also been instrumental, as many options are available to classify and quantify drought severity, from indices based on rainfall alone to those that bring in an evaporation-transpiration component [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. That research has helped create useful drought indexes employed by monitoring, assessment, and worldwide early warning systems. For example, the Standardised Precipitation Index (SPI) and the SPEI are now important for monitoring global and regional drought and supporting decision-making [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Advancements in analyzing and evaluating drought have suggestively increased, mainly in geospatial modeling, simulation, and prediction for future alterations. Scholars recommend spatially distributed machine learning models, like CA-ANN and ANN-Markov Chain, for analyzing and predicting land changes, land surface temperature (LST), and drought changes. These models offer precise capabilities and exceptional behaviors regarding approach and simulation complexities. Neural network models, especially artificial neural networks (ANNs), are mostly used for variation analysis and prediction because they accurately represent complex geographically heterogeneous land-use and land cover alterations. The coupled CA-ANN model, which combines cellular automata with artificial neural networks, effectively simulates land change systems and utilizes \"what-if\" situations in land alteration simulation and modeling. In this study, the MOLUSCE plugin, a user-friendly QGIS plugin compatible with versions 2.00 to 2.99, was utilized to estimate future drought conditions and spatial-temporal transitions using SPI and SPEI. This plugin, introduced by Asia Air Survey (AAS) in 2012, offers various algorithms essential for different techniques required in the study's objectives. The study uses MODIS remote sensing, climate data, and a machine learning approach to analyze the spatio-temporal drought disaster over the Ningxia Hui Autonomous Region of China over the last 20 years. The study aims to recognize the temporal-spatial change of drought, assess the effect of drought on agriculture, and determine the characteristics of drought in Ningxia. The study used multi-temporal MODIS imagery to examine spatiotemporal drought deviations in the study area and assess the correctness of the machine learning (ML) approach in predicting drought. The basic objective of this study is to comprehensively analyze drought patterns in the Ningxia Hui Autonomous Region of China from 2003\u0026ndash;2023.\u003c/p\u003e \u003cp\u003eThus, the main objectives of this study are 1) Identifying spatiotemporal changes of drought using Multi-Temporal MODIS imagery. 2) Examining the impact of drought on agriculture across the Ningxia. 3) Determining the unique characteristics of drought within the specified region. 4) Implementing machine learning approaches to predict drought existence in the future. This study can provide valuable insights into drought dynamics that may inform improved drought management and agricultural planning approaches.\u003c/p\u003e"},{"header":"2 Materials and Methods","content":"\u003cp\u003eThe RS and machine learning techniques (ML) were applied to predict drought deviations in Ningxia. The process utilizes MODIS \u0026amp; Climate Data (2003\u0026ndash;2023) and Precipitation \u0026amp; Evapotranspiration data. Drought and vegetation indices (i.e., KBDI, SPI, SPEI, EVI, and LST) evaluated the variations in drought patterns in the study area. The ANN model was executed to forecast future droughts using SPI and SPEI. The ML component employs a CA-ANN (Convolutional Artificial Neural Network) model, which takes input from the indices SPI (Standardized Precipitation Index) and SPEI (Standardized Precipitation Evapotranspiration Index). The CA-ANN model is used for the future prediction of SPI \u0026amp; SPEI (Drought), ultimately leading to the final map layout as the output of the entire process. The methodology (Fig.\u0026nbsp;\u003cspan\u003e1\u003c/span\u003e) combines remote sensing data, climate data, image processing techniques, and machine learning algorithms to predict and map drought conditions in a specific study area.\u003c/p\u003e\n\u003cdiv id=\"Sec3\"\u003e\n \u003ch2\u003e2.1 Study Area\u003c/h2\u003e\n \u003cp\u003eNingxia is an inland province in northwest China, with a land area of about 66,400 square kilometers and over 6\u0026nbsp;million people (see Fig.\u0026nbsp;\u003cspan\u003e2\u003c/span\u003e). It is characterized by a dry climate, for which the features of scarce rainfall, high evaporation, and significant differences between day and night temperatures are typical. The annual rainfall lies between 200 to 700 mm and decreases gradually from southeast to northwest. The average yearly precipitation is about 300mm, 60\u0026ndash;80% recorded between July and September when East Asian monsoons prevail. The temperature in Ningxia ranges from \u0026minus;\u0026thinsp;10\u0026deg;C to 25\u0026deg;C, generally revealing the typical continental monsoon climate of the North Temperate Zone [\u003cspan\u003e13\u003c/span\u003e]. Its capital is Yinchuan, and its terrain consists mainly of desert, grassland, plains, and mountains, the western Helan Mountains [\u003cspan\u003e14\u003c/span\u003e]. The Yellow River flows through the Province. It is a significant water source for irrigation and hydropower and a cause of problems, including flooding, soil erosion, and desertification.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec4\"\u003e\n \u003ch2\u003e2.2 Datasets\u003c/h2\u003e\n \u003cp\u003eA range of satellites and their sensors were used to capture data for 20 years, as shown in Table\u0026nbsp;\u003cspan\u003e1\u003c/span\u003e. The data was taken in 2003, 2022, 2023. The data sources are Earth Explorer, Earth Engine, Climate Data, and DIVA-GIS. The MOD09GA, MOD11A2, and MCD43A4 products are also derived from MODIS data. These products are generated weekly and have a spatial resolution of 500m-1200 km. The study used climate data from Climate Data to capture long-term trend climate conditions affecting their study area. Climate Data is a networked ecosystem for Climate Data. Climate data is the first step in realizing an overall digital earth. Boundary data DIVA-GIS used to determine the study area focus on the area of interest [\u003cspan\u003e15\u003c/span\u003e].\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 1\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003elists the Datasets used in this study.\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\u003eSatellite/\u003c/p\u003e\n \u003cp\u003eProduct\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSensor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eYear\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpatial Resolution\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSource\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\u003eMODIS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMOD09GA, MOD11A2, MCD43A4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2003\u0026ndash;2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e500m, 1200*1200km, 463.313m\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003ehttps://earthengine.google.com/\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eClimate Data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePrecipitation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2003\u0026ndash;2022\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGround Data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003ehttps://en.climate-data.org/asia/china-110/\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBoundary\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2023\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003ehttps://www.diva-gis.org/gdata\u003c/span\u003e\u003c/span\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=\"Sec5\"\u003e\n \u003ch2\u003e2.3 Methods\u003c/h2\u003e\n \u003cdiv id=\"Sec6\"\u003e\n \u003ch2\u003e2.3.1 Drought Indices\u003c/h2\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab2\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 2\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eThe summary of vegetation and drought indices.\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\u003eProduct\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eData Used\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpatial Resolution\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTemporal Resolution\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eFormula\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\u003eLST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMOD11A2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1*1 km\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8-day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:LST=DN\\times\\:0.02-273.15\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEVI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMCD43A4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e463.313m\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e16-Day\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:EVI=2.5\\text{*}\\frac{NIR-Red}{(NIR+C1\\text{*}Red-C2\\text{*}BLUE+L)}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCHIRPS precipitation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.05\u0026deg;\u0026times;0.05\u0026deg;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMonthly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:=\\frac{X-Xm}{\\sigma\\:}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPEI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCSIC/SPEI/2_8\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.5\u0026deg;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMonthly\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:SPEI=W-\\frac{c0+\\text{c}1\\text{W}+\\text{c}2{W}^{2}}{1+d1W+d2{W}^{2}+d3{W}^{3}}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eKBDI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWTLAB/KBDI/v1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4*4km\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDaily\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:\\text{K}\\text{B}\\text{D}\\text{I}=\\text{Q}+\\frac{\\left(800-Q\\right).\\left(0.968.{e}^{0.0486.T\\:}-8.30\\right).\\varDelta\\:t}{1+10.88.{e}^{0.0486.T\\:}}\\:.\\:{10}^{-3}\\)\u003c/span\u003e\u003c/span\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 \u003cp\u003eSummary of the vegetation and drought indices is given in the Table\u0026nbsp;\u003cspan\u003e2\u003c/span\u003e along with the formula of each index. Spatial and temporal resolution of indices are different as mentioned in the table.\u003c/p\u003e\n \u003cp\u003eThe Enhanced Vegetation Index (EVI) is one of the most commonly used RS indexes for estimating vegetation cover. It is measured using the reflectance values of near-infrared (NIR), red, and blue light from the Earth\u0026apos;s surface [\u003cspan\u003e16\u003c/span\u003e]. In this study area, research was conducted on the MODIS/MCD43A4 surface reflectance variables on Google Earth Engine (GEE) using EVI. Its ranges typically remain from \u0026minus;\u0026thinsp;1.0 to 1.0, and higher values show more vegetation. By combining the blue band into the calculation, EVI is less sensitive to atmospheric effects, and its ability to differentiate variations in vegetation cover is improved.\u003c/p\u003e\n \u003cdiv id=\"Equ1\"\u003e\n \u003cdiv id=\"FileID_Equ1\" name=\"EquationSource\"\u003e$$\\:EVI=2.5\\text{*}\\frac{NIR-Red}{(NIR+C1\\text{*}Red-C2\\text{*}BLUE+L)}$$\u003c/div\u003e\n \u003cdiv\u003e1\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eLST is an essential parameter for observing the surface temperature of the Earth and its variations over time. For LST on GEE, the MODIS/061/MOD11A2 dataset was used in the study area giving an average of 8-day 1x1 km LST products.\u003c/p\u003e\n \u003cdiv id=\"Equ2\"\u003e\n \u003cdiv id=\"FileID_Equ2\" name=\"EquationSource\"\u003e$$\\:LST=DN\\times\\:0.02-273.15$$\u003c/div\u003e\n \u003cdiv\u003e2\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eML algorithms find associations and patterns in data, permitting software to recover its performance over time [\u003cspan\u003e17\u003c/span\u003e]. The standardized precipitation index (SPI) is widely used to characterize agricultural drought across different timescales [\u003cspan\u003e18\u003c/span\u003e]. SPI was computed in the study area of Ningxia by using CHIRPS precipitation data. The script was accomplished through GEE, and the work was split into SPI (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12) calculations. The first calculation was based on the \u0026quot;common\u0026quot; SPI, computed on an n-month basis. A SPI computed on one month would then be referred to as \u0026quot;SPI-1\u0026quot;, on six months, \u0026quot;SPI-6,\u0026quot; and so on [\u003cspan\u003e19\u003c/span\u003e]. The second SPI was computed based on dates of MODIS captures to illustrate changes in precipitation and thus in how the Earth\u0026apos;s precipitation is affected by the changes of climate that the globe is currently undergoing [\u003cspan\u003e20\u003c/span\u003e]. SPI classification category is given in Table\u0026nbsp;\u003cspan\u003e3\u003c/span\u003e.\u003c/p\u003e\n \u003cdiv id=\"Equ3\"\u003e\n \u003cdiv id=\"FileID_Equ3\" name=\"EquationSource\"\u003e$$\\:SPI=\\:-\\left(t-\\:\\frac{{c}_{0}+{c}_{1}t+{c}_{2}{t}^{2}}{1+{d}_{1}t+{d}_{2}{t}^{2}+{d}_{3}{t}^{3}}\\right)$$\u003c/div\u003e\n \u003cdiv\u003e3\u003c/div\u003e\n \u003c/div\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 3\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003elists SPI Classification [\u003cspan\u003e21\u003c/span\u003e]\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSPI Category\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eValue\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\u003eLess than \u0026minus;\u0026thinsp;2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eExtremely dry\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween \u0026minus;\u0026thinsp;1.5 \u0026amp; -2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSeverely dry\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween \u0026minus;\u0026thinsp;1 \u0026amp; -1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDry\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween \u0026minus;\u0026thinsp;0.5 \u0026amp; -1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerately dry\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween 0.5 \u0026amp; -0.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNormal\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween 0.5 \u0026amp; 1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween 1 \u0026amp; 1.5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerately wet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBetween 1.5 \u0026amp; 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSeverely wet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMore than 2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eExtremely wet\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eSPEI is an index based on precipitation and evapotranspiration data, recognized as the Standardised Precipitation Evapotranspiration Index (SPEI) [\u003cspan\u003e22\u003c/span\u003e], used for the study area, Ningxia. The CSIC/SPEI dataset gives global SPEI data for the entire Earth at a spatial resolution of 0.5\u0026ordm; [\u003cspan\u003e23\u003c/span\u003e]. A 1-month, 3-month, 6-month, 9-month, and 24-month monitoring was done in the study area, Ningxia. The dataset monitored drought fluctuations (wet and dry spells) from 2003 to 2023 in the study area. Different categories of SPEI are classified as discussed in Table\u0026nbsp;\u003cspan\u003e4\u003c/span\u003e.\u003c/p\u003e\n \u003cdiv id=\"Equ4\"\u003e\n \u003cdiv id=\"FileID_Equ4\" name=\"EquationSource\"\u003e$$\\:SPEI=W-\\frac{{c}_{0}+{c}_{1}\\text{W}+{c}_{2}{W}^{2}}{1+{d}_{1}W+{d}_{2}{W}^{2}+{d}_{3}{W}^{3}}$$\u003c/div\u003e\n \u003cdiv\u003e4\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eWhere:\u003c/p\u003e\n \u003cp\u003e\u003cspan\u003e\u0026nbsp;\u003cspan\u003e\\(\\:W=\\:\\sqrt{-2\\text{ln}\\left(P\\right)\\:\\:}\\)\u003c/span\u003e\u0026nbsp;\u003c/span\u003e for P\u0026thinsp;\u0026le;\u0026thinsp;0.5 (5)\u003c/p\u003e\n \u003cp\u003eP\u0026thinsp;=\u0026thinsp;probability of exceeding a determined D value, \u003cspan\u003e\u003cspan\u003e\\(\\:p=1-f\\left(x\\right);\\)\u003c/span\u003e\u003c/span\u003e When P\u0026thinsp;\u0026gt;\u0026thinsp;0.5, \u003cspan\u003e\u003cspan\u003e\\(\\:p=1-P,\\)\u003c/span\u003e\u003c/span\u003e constants are:\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab4\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003e\u003cbr\u003e\u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:{c}_{0}\\:=\\:2.515517\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:{d}_{1}\\:=\\:1.432788\\)\u003c/span\u003e\u003c/span\u003e\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\u003cspan\u003e\u003cspan\u003e\\(\\:\\:{c}_{1}\\:=\\:0.802853\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:{d}_{2}\\:=\\:0.189269\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:\\:{c}_{2}=\\:0.010328\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cspan\u003e\u003cspan\u003e\\(\\:{d}_{3}=\\:0.001308\\)\u003c/span\u003e\u003c/span\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 \u003cdiv\u003e\n \u003cdiv align=\"left\"\u003e\u003cstrong\u003eTable 4.\u0026nbsp;\u003c/strong\u003eSPEI Classification (Fu et al. 2022).\u003c/div\u003e\n \u003ctable id=\"Taba\" border=\"1\"\u003e\n \u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSPEI Category\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eValue\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\u003eExtremely wet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMore than 2.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eVery wet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.50 to 1.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerately wet\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 to 1.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNear Normal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-0.99 to 0.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eModerately dry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-1.00 to -1.49\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSeverely dry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-1.50 to -1.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eExtremely dry\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLess than \u0026minus;\u0026thinsp;2.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eDaily maximum temperature and precipitation measurements determine the Keetch-Byram drought index (KBDI) that bears on evapotranspiration [\u003cspan\u003e24\u003c/span\u003e]. The daily maximum temperature enables calculating the amount of water evaporated from the soil surface evaporative demand and total precipitation, from which KBDI, the soil moisture deficit, and the amount needed to bring a site\u0026apos;s soil moisture to field capacity can be determined. The UTOKYO/WTLAB/KBDI/v1 dataset, which provides a continuous reference scale that divides the moisture regime of the soil and duff layers into eight classes from 0.0 (no moisture deficit) to 800.0 (extreme drought) by the cumulative drying that occurs during each day of no rain (Table\u0026nbsp;\u003cspan\u003e5\u003c/span\u003e), the rate of which, in turn, depends on the daily highs [\u003cspan\u003e25\u003c/span\u003e]. KBDI values provide a relative measure of soil moisture and fire risk, making them valuable indicators in assessing the impact of drought on potential wildfire hazards.\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab5\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 5\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003eLists KBDI Classification [\u003cspan\u003e26\u003c/span\u003e].\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"2\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eKBDI Value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCategory\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\u003e0 to 200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIndicates high soil moisture, suggesting a lower risk of wildfire in the presence of ample water content.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e200 to 400\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIt represents moderate soil moisture, signifying a moderate wildfire risk, especially in regions experiencing drought.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e400 to 600\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eReflects low soil moisture, indicating an elevated wildfire risk, particularly in drought areas.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e600 to 800\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIt signifies deficient soil moisture, highlighting an extreme wildfire risk, particularly in regions undergoing severe drought.\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=\"Equ5\"\u003e\n \u003cdiv id=\"FileID_Equ5\" name=\"EquationSource\"\u003e$$\\:\\text{K}\\text{B}\\text{D}\\text{I}\\:=\\text{Q}+\\frac{\\left(800-Q\\right).\\left(0.968.{e}^{0.0486.T\\:}-8.30\\right).\\varDelta\\:t}{1+10.88.{e}^{0.0486.T\\:}}\\:.\\:{10}^{-3}$$\u003c/div\u003e\n \u003cdiv\u003e6\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eQ, which represents the previous day\u0026apos;s KBDI adjusted by the net rainfall in inches per hundred (cf. details below); T, the air temperature in degrees Fahrenheit; \u0026Delta;t, the time increase (typically one day); and P, signifying the mean annual precipitation in inches.\u003c/p\u003e\n \u003cdiv id=\"Equ6\"\u003e\n \u003cdiv id=\"FileID_Equ6\" name=\"EquationSource\"\u003e$$\\:Q={KBDI}_{t-1\\:}-\\:\\text{P}{net}_{t}\\:.\\:100$$\u003c/div\u003e\n \u003cdiv\u003e7\u003c/div\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Equ7\"\u003e\n \u003cdiv id=\"FileID_Equ7\" name=\"EquationSource\"\u003e$$\\:\\text{P}{net}_{t}=\\text{m}\\text{a}\\text{x}[0,{P}_{t\\:}-\\text{m}\\text{a}\\text{x}(0,{P}_{\\text{l}\\text{i}\\text{m}\\:}-{\\sum\\:}_{i=1}^{rr-1}{P}_{\\text{t}-\\text{i}\\:})$$\u003c/div\u003e\n \u003cdiv\u003e8\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eWith \u0026quot;rr\u0026quot; denoting the count of consecutive days on which rain has occurred.\u003c/p\u003e\n \u003cp\u003eThe agricultural drought disaster model summary using the machine learning model (MLM) is in Table\u0026nbsp;\u003cspan\u003e6\u003c/span\u003e. Various MLMs are used for drought prediction, depending on the available data. For example, meteorological data and a combination of CART and SVM models are used to predict SPEI annually. In contrast, CART Cubist models can use MODIS data to predict SPI on a seasonal basis, including early, growing, middle, and late seasons. LSTM models can predict SPEI for up to 12 months based on soil moisture, LST, ET, EVI, and precipitation [\u003cspan\u003e27\u003c/span\u003e].\u003c/p\u003e\n \u003cdiv\u003e\n \u003ctable id=\"Tab6\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv\u003eTable 6\u003c/div\u003e\n \u003cdiv\u003e\n \u003cp\u003elists the Agricultural drought prediction summary using machine learning models (MLMs).\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"6\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eData type\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eForecaster variables\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eResponse variable\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003ePredicting lead time\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eOutcome\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\u003eCART and SVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMODIS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEVI, NDVI, LST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPEI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSeasonal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIncreased drought area prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMeteorological data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSlope, aspect, elevation, annual precipitation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAgricultural drought prediction\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCubist\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMODIS, TRMM and climate data\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEVI, LST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSeasonal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSevere Drought Index Mapping\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSVM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSoil Moisture\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLST, ET, EVI, precipitation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSPEI and crop yield\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12-months\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eDrought severity distribution maps\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=\"Sec7\"\u003e\n \u003ch2\u003e2.3.2 Change Evaluation and Future Prediction using SPI and SPEI\u003c/h2\u003e\n \u003cp\u003eIn this study, the MOLUSCE plugin in QGIS was employed to simulate the SPI and SPEI change between their classes (extremely dry, severely dry, moderately dry, normal, and wet) and estimate spatiotemporal changes of drought for the periods 2003 to 2023. SPI (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12) and SPEI (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12) maps of each year were produced. The 2003 and 2023 drought (SPI and SPEI) created area variation and transition probability matrixes. The artificial neural network (ANN) multilayer perception strategy was implemented. Precipitation, temperature, and PET were taken as the determinant factors for future drought variation prediction. In drought change analysis and prediction, these variables are frequently used because they deliver verifiable information on the effect of anthropogenic and natural factors on SPI and SPEI variations.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e"},{"header":"3 Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Enhanced Vegetation Index (EVI) changes from 2003\u0026ndash;2023\u003c/h2\u003e \u003cp\u003eThe maximum EVI in 2003 did not exceed 0.38, whereas the maximum in 2023 was 0.33, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The maps in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e show that EVI in Ningxia predominantly decreased over the past 20 years. Climate change has caused a significant incline in the frequency and severity of drought. This puts the region at greater risk of soil erosion. 2003 provides an early baseline for pre-drought assessment, while 2018 and 2023 show more recent conditions after prolonged drought. The selected years deliver good temporal sampling aligned with significant droughts. These are all valuable factors for generating a robust EVI trend analysis for assessing desertification. Machine learning has been used to study desertification change in Ningxia. From 2003 to 2023, the average annual soil erosion rate in Ningxia, China, increased [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Land Surface Temperature (LST) changes from 2003\u0026ndash;2023\u003c/h2\u003e \u003cp\u003eThe map below shows the Land Surface Temperature (LST) of Ningxia, China, from 2003 to 2023. The LST is in degrees Celsius (\u0026deg;C). The LST is 4.6\u0026deg;C to 31\u0026deg;C, as displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. The LST of Ningxia has been increasing with time. In 2003, the average LST of Ningxia was 14.3\u0026deg;C. The average LST of Ningxia will be 16.7\u0026deg;C by 2023. The inclination of the trend line in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e shows the increase in the temperature from 31\u0026deg;C to 34\u0026deg;C from 2003 to 2023. The years were likely chosen to provide evenly spaced data points over the 20 years to see the overall trend in LST over time. 2003 provides an early baseline temperature, while 2006, 2012, 2015, 2018, and 2023 allow for comparison of the LST approximately every 3, 6, 9, and 15 years after the initial 2003 data point.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Keetch-Byram Drought Index (KBDI) changes from 2003\u0026ndash;2023\u003c/h2\u003e \u003cp\u003eKBDI is a measure used to assess the dryness of an area and the potential risk of wildfire. In Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e6\u003c/span\u003e, the spatial distribution of KBDI in Ningxia changes over time. In 2007, the KBDI of Ningxia was 580; by 2023, the average KBDI of Ningxia had increased to 680. This increase in KBDI indicates that Ningxia has become drier over time and that the risk of wildfires has increased. Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003e shows that drought tends to grow from 2007 to 2023.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Standardized Precipitation Index (SPI) changes from 2003\u0026ndash;2023\u003c/h2\u003e \u003cp\u003eNingxia experienced various wet and dry conditions over the 2003 to 2023 period, as illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e8\u003c/span\u003e. Most of the years experienced drought in the north part of the region in 2003, 2005, 2013, 2014, 2017, 2019, and 2023. However, the opposite pattern occurred in 2004, 2007, 2009, 2011, 2012, 2016, and 2022, which shows the extreme drought experienced in the southern part of Ningxia.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSPI-1, SPI-3, SPI-6, SPI-9, and SPI-12 variations of the drought pattern in Ningxia from 2003\u0026ndash;2023 are displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e9\u003c/span\u003e. The timescales of SPI determine the wet changes, and the negative (decrease) in the timescales demonstrates the dry spells. In SPI-1 2003, the dry spell ranges from 0.7 to -1, and the wet spell is from 1 to 1.3. 2014, 2018, 2021 and 2022. SPI-3 has a high damp spell in 2013 and 2018, while dry spells with \u0026minus;\u0026thinsp;1.5 to -2 are in 2003, 2011,2021, and 2023. SPI-6 has more dry spell fluctuations, as in 2006, 2007, and 2022, and SPI is between \u0026minus;\u0026thinsp;2 to -2.4. The wet spell of SPI-9 reached\u0026thinsp;+\u0026thinsp;2 to +\u0026thinsp;3 in 2015 and 2018, while the highest dry spell was \u0026minus;\u0026thinsp;1.5 in 2021. SPI-12 demonstrates more dry spells in 2005, 2006, 2011, 2015, 2019, 2021, and 2022.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Standardized Precipitation Evapotranspiration Index (SPEI)\u003c/h2\u003e \u003cp\u003eIn 2003, most of the region experienced normal conditions. In 2004, 2005, 2015, 2018, and 2019, the SPEI map shows that the northeastern half of Ningxia experienced extremely dry conditions, while in 2006, the northern half of the area experienced extreme conditions. Overall, more drought years were 2004, 2005, 2006, 2009, 2015, 2016, 2018, 2019, and 2021.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12 changes in the drought pattern in Ningxia from 2003\u0026ndash;2023 are displayed in Fig.\u0026nbsp;\u003cspan refid=\"Fig12\" class=\"InternalRef\"\u003e11\u003c/span\u003e. Different timescales of SPEI determine the variations in the wet and dry spells in Ningxia. In SPEI-1, the damp spell ranges from 1 to 2 in 2007, 2008, 2016, and 2018, while the dry spell is between \u0026minus;\u0026thinsp;0.7 to -2.8 in 2004, 2009, 2015, and 2021. SPEI-3 had the highest wet spell in 2012 and the highest dry spell in 2015. SPEI-6 has frequent dry spells in 2004, 2005, 2006, 2009, 2015, 2020, and 2023. The 9-month scale (SPEI-9) reflects the long-term drought process with more extreme droughts in 2003, 2005, 2006, 2009, 2011, 2013, and 2015. The 12-month scale (SPEI-12) of SPEI reflects more frequent extreme droughts in 2005, 2006, 2009,2011, 2015, 2020, 2022, and 2023.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e3.6 Drought Change and Prediction Using the CA-ANN Model\u003c/h2\u003e \u003cp\u003eThe future prediction, along with the previous (2003, 2013) and recent (2023) map of SPI in Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e12\u003c/span\u003e, shows predicted dryness or drought conditions for Ningxia in the year 2033 in (a4, b4, c4, d4, and e4). The map uses the Standardised Precipitation Index (SPI) to predict drought conditions. The map of SPI (SPI-1, SPI-3, SPI-6, SPI-9, SPI-12) indicates that the areas at risk of drought in 2033 are northwest Ningxia, the central Plains, and the northeast. SPI-12 shows more frequent severe and extreme drought in the upper north and east part of the region for 2033.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe future prediction result in Fig.\u0026nbsp;\u003cspan refid=\"Fig14\" class=\"InternalRef\"\u003e13\u003c/span\u003e of SPEI indicates predicted drought conditions for 2033 in (a4, b4, c4, d4, e4). The map uses the SPEI (SPEI-1, SPEI-3, SPEI-6, SPEI-9, SPEI-12) to predict drought conditions. According to future predictions, the areas at risk of drought in (a) SPEI-1 are the north and western Ningxia in (a1) and (a4). SPEI-1 of 2013 in (a2) had an extreme drought in the north region, while the southern region of Ningxia is moderately dry. The (b) SPEI-3 and (c) SPEI-6 display the northern and southern regions of Ningxia to be extremely dry in (b4), and in (c4), extreme drought is predicted to be in the western region of Ningxia. SPEI-9 (d4) shows Ningxia's northeastern and southwestern regions to be severely dry. In SPEI-12, the (e4) is expected to be extremely dry in the western and southern parts in 2033.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe kappa coefficient for SPI and SPEI timescale predicted results employing the CA-ANN Model is given in Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e. The results show that the SPEI-6 had the highest degree of agreement (96%), and the SPEI-9 had the lowest degree of agreement (65%).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eFuture Prediction for SPI and SPEI.\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\" morerows=\"1\" rowspan=\"2\"\u003e\u0026nbsp;\u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCurrent\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePredicted\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eKappa Value for 2033\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2023\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2033\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCA-ANN Model\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eValidation\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSPI Timescale\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPI-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.835\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.82965\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.93093\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPI-3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.713\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.355\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.81178\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.93639\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPI-6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.613\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.475\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.93243\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.95226\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPI-9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-0.555\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.578\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.86640\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.95697\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPI-12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.320\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.991\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.91596\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.95360\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"5\" nameend=\"c5\" namest=\"c1\"\u003e \u003cp\u003e\u003cb\u003eSPEI Timescale\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPEI-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.898\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.394\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.93085\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.86665\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPEI-3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.127\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.365\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.88649\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.81905\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPEI-6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.622\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.827\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.83258\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.96389\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPEI-9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-1.369\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-1.735\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.73214\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.65756\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSPEI-12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-2.046\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-2.894\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.92242\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.88763\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"},{"header":"4 Discussion","content":"\u003cp\u003eProperly monitoring drought has significant implications for mapping drought spatial anomalies over time, risk management, drought preparedness, and agricultural irrigation for the future [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e]. This study also assessed the changes in drought patterns yearly from 2003 to 2023. In this research, remote sensing and machine learning approaches are used to get effective drought monitoring results, which devastates agriculture and threatens the food security of China's Ningxia Hui Autonomous Region. Remotely sensed vegetation and drought monitoring combined with machine learning to understand unprecedented drought intensity patterns across Ningxia during the study period. An Enhanced Vegetation Index (EVI) was performed to monitor vegetation anomalies over time. Land Surface Temperature (LST) is estimated to analyze the correlation of temperature with drought. Standardized Precipitation Index (SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12), Standardized Precipitation Evapotranspiration Index (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12), and Keetch-Byram Drought Index (KBDI) were used as drought indicators for drought monitoring across study area (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe results of EVI quantify the impacts of vegetation on health, which showed that 2003 experienced the highest vegetation, ranging from 0.27 to 0.38. In contrast, the range of vegetation in 2023 was 0.24 to 0.33, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. The results of EVI claim that vegetation has declined in Ningxia over the previous two decades. The assessment of LST indicated that in 2003, the highest temperature was about 31\u0026deg;C, which increased to 38\u0026deg;C in 2023. The increase in temperature was about 4\u0026deg;C in the two decades. Fi 4 displays the increase in land surface temperature of Ningxia from 2003 to 2023. The rise in temperature is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e5\u003c/span\u003e, where the trend line determines the inclination in the temperature. EVI and LST show an inverse relation with each other. Our findings suggest that the increased LST can be due to decreased vegetation in the Ningxia. [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e] also investigated how increased temperature reduces vegetation and vice versa. The index KBDI indicates drought intensities from soil moisture deficits [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. Multi-sensor and multi-index spatial analyses delineate drought risks spatially and intensity-wise heterogeneous nature across Ningxia. At the same time, most of Ningxia reveals moderate to severe drought conditions, the highest drought intensity over most indicators clustered centrally. KBDI flags this region as having the most severe and agriculturally impactful drought severity, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e6\u003c/span\u003e and Fig.\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e7\u003c/span\u003e; soils dry out from water scarcity in the area historically.\u003c/p\u003e \u003cp\u003eFurthermore, the results of SPI illustrated that the overall SPI noticeably decreased from 2003 to 2023. The SPI indicated that the most affected years were 2005, 2006, 2009, 2012, 2016, 2020, and 2022, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e8\u003c/span\u003e. SPEI indicated the study area has become drier in several years, and Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e10\u003c/span\u003e shows years of severe and extreme droughts in the years 2004, 2005, 2006, 2009, 2015, 2016, 2018, 2019, and 2021. SPI and SPEI quantify drought from precipitation deficits [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOur study found the most robust relationship between SPEI and KBDI, suggesting that in addition to precipitation deficits by SPEI, capturing the evapotranspiration dynamics gave a critical indication that dry enough soil moisture affects agriculture. This relationship was also found in the relevant study by (Thanh ) in 2023. EVI has weaker relationships, indicating the vegetation response to drought depends on the specific vegetation species, farming practices, and environmental micro-climates [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Moreover, the study suggested that SPEI was the most efficient indicator and gave better results than SPI and KBDI. This indicator is also considered superior in the study conducted by [\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe machine learning algorithm integrated multi-index datasets. CA-ANN model provides a more detailed and robust agricultural drought intensity zoning across the spatiotemporal scales of Ningxia. These algorithms are used because of their simplicity, effectiveness, and reliability. The algorithm identified central Ningxia as a zone of recurring high drought severity, again showing the strength of the spatial monitoring that comes from integrating multiple indices. In addition to that, Fig.\u0026nbsp;\u003cspan refid=\"Fig13\" class=\"InternalRef\"\u003e12\u003c/span\u003e shows the comparison of the previous and recent decades with the future prediction of drought conditions for the year 2033 using SPI-1, SPI-3, SPI-6, SPI-9, and SPI-12. It shows that northwest and northeastern Ningxia are at risk of facing extreme drought conditions. SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12 for 2033 also show that some areas of Ningxia are at risk of drought. These areas are north, west, south, northeastern, southwestern, and northwestern parts of Ningxia, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig14\" class=\"InternalRef\"\u003e13\u003c/span\u003e. Thus, remote sensing and machine learning can be used for drought monitoring and assessing its impacts on the relevant [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. In future studies, testing group performance can be evaluated by integrating the Root Mean Square Error (RMSE) technique. Future predictions of drought can also help in drought preparedness and recovery.\u003c/p\u003e"},{"header":"5 Conclusion","content":"\u003cp\u003eThis work evaluated the temporal-spatial variation of drought in Ningxia Hui Autonomous Region, China. The results show an increase in severe and moderate drought areas of 15% and 10%, while there was a 25% decrease in mild drought areas in the 2003\u0026ndash;2023. These results can guide the region's direction regarding drought mitigation and adaptation strategies. The study area is a critical food production area in northwest China that is severely affected by drought. MODIS reflectance data, LST, precipitation, evapotranspiration data, and meteorological data from 14 meteorological stations were used to assess the temporal-spatial variation of drought. The results showed that applying drought indices and using machine learning, like the CA-ANN model, represents a comprehensive utilization approach for discerning and categorizing drought severity levels. This study is significant given the acceleration of severe and moderate drought areas and the decline of mild drought areas in the study area during this two-decade timeframe. This spatial and time-based reasoning is necessary to inform the region's policy and decision-makers responsible for drought management and agricultural activities. This approach and its findings add to the required directions to monitor and mitigate the influence of drought on the environment and economy of the Ningxia Hui Autonomous Region.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eM.A.K., data curation, investigation, software, code, writing\u0026mdash;original draft; J.Z., conceptualization, funding acquisition, supervision; writing\u0026mdash;review; S.A., software, visualization, writing\u0026mdash;review; Z.Z., code, visualization, writing\u0026mdash;review; H.U., visualization, writing\u0026mdash;review; A.G., writing\u0026mdash;review. All authors have read and agreed to the published version of the manuscript\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eAcknowledgements.\u0026nbsp;\u003c/strong\u003eThis work was \u0026nbsp; supported by the Central Guiding Local Science and Technology Development Fund of Shandong-Yellow River Basin Collaborative Science and Technology Innovation Special Project (No. YDZX2023019).\u003c/p\u003e\n"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eG. Zhang, X. Su, O. O. Ayantobo, and K. Feng, \u0026ldquo;Drought monitoring and evaluation using ESA CCI and GLDAS-Noah soil moisture datasets across China,\u0026rdquo; Theor Appl Climatol, vol. 144, no. 3, pp. 1407\u0026ndash;1418, 2021, doi: 10.1007/s00704-021-03609-w.\u003c/li\u003e\n\u003cli\u003eQ. Zhang et al., \u0026ldquo;Causes and Changes of Drought in China: Research Progress and Prospects,\u0026rdquo; Journal of Meteorological Research, vol. 34, no. 3, pp. 460\u0026ndash;481, 2020, doi: 10.1007/s13351-020-9829-8.\u003c/li\u003e\n\u003cli\u003eY. Xu et al., \u0026ldquo;Drought assessment of China in 2002\u0026ndash;2017 based on a comprehensive drought index,\u0026rdquo; Agric For Meteorol, vol. 319, p. 108922, 2022, doi: https://doi.org/10.1016/j.agrformet.2022.108922.\u003c/li\u003e\n\u003cli\u003eP. Sun, Z. Ma, Q. Zhang, V. P. Singh, and C.-Y. Xu, \u0026ldquo;Modified drought severity index: Model improvement and its application in drought monitoring in China,\u0026rdquo; J Hydrol (Amst), vol. 612, p. 128097, 2022, doi: https://doi.org/10.1016/j.jhydrol.2022.128097.\u003c/li\u003e\n\u003cli\u003eW. Liu, F. Sun, S. Sun, L. Guo, H. Wang, and H. Cui, \u0026ldquo;Multi-scale assessment of eco-hydrological resilience to drought in China over the last three decades,\u0026rdquo; Science of The Total Environment, vol. 672, pp. 201\u0026ndash;211, 2019, doi: https://doi.org/10.1016/j.scitotenv.2019.03.408.\u003c/li\u003e\n\u003cli\u003eQ. Zhang et al., \u0026ldquo;Droughts across China: Drought factors, prediction and impacts,\u0026rdquo; Science of The Total Environment, vol. 803, p. 150018, 2022, doi: https://doi.org/10.1016/j.scitotenv.2021.150018.\u003c/li\u003e\n\u003cli\u003eY. Wang, W. Zhao, Q. Zhang, and Y. Yao, \u0026ldquo;Characteristics of drought vulnerability for maize in the eastern part of Northwest China,\u0026rdquo; Sci Rep, vol. 9, no. 1, p. 964, 2019, doi: 10.1038/s41598-018-37362-4.\u003c/li\u003e\n\u003cli\u003eL. I. U. Bin, P. Li, Z. Liping, and R. E. N. Shuyong, \u0026ldquo;Drought Resistance Study of 10 Major Ornamental Shrub in Ningxia,\u0026rdquo; Acta Botanica Boreali-Occidentalia Sinica, vol. 33, no. 9, pp. 1808\u0026ndash;1816, [Online]. Available: xbzwxben/article/abstract/20130914\u003c/li\u003e\n\u003cli\u003eD. Xu, Y. Ding, H. Liu, Q. Zhang, and D. Zhang, \u0026ldquo;Applicability of a CEEMD\u0026amp;ndash;ARIMA Combined Model for Drought Forecasting: A Case Study in the Ningxia Hui Autonomous Region,\u0026rdquo; Atmosphere (Basel), vol. 13, no. 7, 2022, doi: 10.3390/atmos13071109.\u003c/li\u003e\n\u003cli\u003eC. Wang et al., \u0026ldquo;Impacts of drought on maize and soybean production in northeast China during the past five decades,\u0026rdquo; Int J Environ Res Public Health, vol. 17, no. 7, Apr. 2020, doi: 10.3390/ijerph17072459.\u003c/li\u003e\n\u003cli\u003eC. Tan, J. Yang, X. Wang, D. Qin, B. Huang, and H. Chen, \u0026ldquo;Drought disaster risks under CMIP5 RCP scenarios in Ningxia Hui Autonomous Region, China,\u0026rdquo; Natural Hazards, vol. 100, no. 3, pp. 909\u0026ndash;931, 2020, doi: 10.1007/s11069-019-03811-0.\u003c/li\u003e\n\u003cli\u003eT. Wu, B. Li, L. Lian, Y. Zhu, and Y. Chen, \u0026ldquo;Assessment of the Combined Risk of Drought and High-Temperature Heat Wave Events in the North China Plain during Summer,\u0026rdquo; Remote Sens (Basel), vol. 14, no. 18, Sep. 2022, doi: 10.3390/rs14184588.\u003c/li\u003e\n\u003cli\u003eZ. Liu et al., \u0026ldquo;An Analysis of the Spatiotemporal Variation in Wind Erosion\u0026ndash;Climate Erosion Force in the Ningxia Region and Its Driving Factors,\u0026rdquo; Atmosphere (Basel), vol. 14, no. 6, Jun. 2023, doi: 10.3390/atmos14060963.\u003c/li\u003e\n\u003cli\u003eY. ZHENG, H. MENG, X. XIE, and S. SHI, \u0026ldquo;Dynamics of Climate Change, Drought and Migration: A Mixed Method Research in Ningxia,\u0026rdquo; Chinese Journal of Urban and Environmental Studies, vol. 05, no. 03, p. 1750021, Sep. 2017, doi: 10.1142/s234574811750021x.\u003c/li\u003e\n\u003cli\u003eX. Zhao, H. Xia, B. Liu, and W. Jiao, \u0026ldquo;Spatiotemporal Comparison of Drought in Shaanxi\u0026amp;ndash;Gansu\u0026amp;ndash;Ningxia from 2003 to 2020 Using Various Drought Indices in Google Earth Engine,\u0026rdquo; Remote Sens (Basel), vol. 14, no. 7, 2022, doi: 10.3390/rs14071570.\u003c/li\u003e\n\u003cli\u003eR. Singh, M. Kumari, S. Bindal, and I. Gupta, \u0026ldquo;Quantification of Drought Condition Using Drought Indices: A Review,\u0026rdquo; in Climate Change and Water Security, S. Kolathayar, A. Mondal, and S. C. Chian, Eds., Singapore: Springer Singapore, 2022, pp. 231\u0026ndash;241.\u003c/li\u003e\n\u003cli\u003eY. Zhao et al., \u0026ldquo;Drought Monitoring and Performance Evaluation Based on Machine Learning Fusion of Multi-Source Remote Sensing Drought Factors,\u0026rdquo; Remote Sens (Basel), vol. 14, no. 24, Dec. 2022, doi: 10.3390/rs14246398.\u003c/li\u003e\n\u003cli\u003eZ. Pei, S. Fang, L. Wang, and W. Yang, \u0026ldquo;Comparative analysis of drought indicated by the SPI and SPEI at various timescales in inner Mongolia, China,\u0026rdquo; Water (Switzerland), vol. 12, no. 7, Jul. 2020, doi: 10.3390/w12071925.\u003c/li\u003e\n\u003cli\u003eQ. Zhang, J. Li, V. P. Singh, and Y. Bai, \u0026ldquo;SPI-based evaluation of drought events in Xinjiang, China,\u0026rdquo; Natural Hazards, vol. 64, no. 1, pp. 481\u0026ndash;492, 2012, doi: 10.1007/s11069-012-0251-0.\u003c/li\u003e\n\u003cli\u003eQ. He, M. Wang, K. Liu, B. Li, and Z. Jiang, \u0026ldquo;Spatiotemporal analysis of meteorological drought across China based on the high-spatial-resolution multiscale SPI generated by machine learning,\u0026rdquo; Weather Clim Extrem, vol. 40, p. 100567, 2023, doi: https://doi.org/10.1016/j.wace.2023.100567.\u003c/li\u003e\n\u003cli\u003eC. Tan, J. Yang, and M. Li, \u0026ldquo;Temporal-spatial variation of drought indicated by SPI and SPEI in Ningxia Hui Autonomous Region, China,\u0026rdquo; Atmosphere (Basel), vol. 6, no. 10, pp. 1399\u0026ndash;1421, Oct. 2015, doi: 10.3390/atmos6101399.\u003c/li\u003e\n\u003cli\u003eH. Zhao, G. Gao, W. An, X. Zou, H. Li, and M. Hou, \u0026ldquo;Timescale differences between SC-PDSI and SPEI for drought monitoring in China,\u0026rdquo; Physics and Chemistry of the Earth, vol. 102, pp. 48\u0026ndash;58, Dec. 2017, doi: 10.1016/j.pce.2015.10.022.\u003c/li\u003e\n\u003cli\u003eC. Hao, J. Zhang, and F. Yao, \u0026ldquo;Combination of multi-sensor remote sensing data for drought monitoring over Southwest China,\u0026rdquo; International Journal of Applied Earth Observation and Geoinformation, vol. 35, no. PB, pp. 270\u0026ndash;283, 2015, doi: 10.1016/j.jag.2014.09.011.\u003c/li\u003e\n\u003cli\u003eW. Takeuchi et al., \u0026ldquo;Indonesia 4 Institute of Meteorology, Hydrology and Environment (IMHEN), Vietnam 5 Myanmar Geospatial and Resource Program (MGRP),\u0026rdquo; MRC, 2017. [Online]. Available: http://sharaku.eorc.jaxa.jp/GSMaP\u003c/li\u003e\n\u003cli\u003eM. H\u0026auml;usler, J. P. Nunes, J. M. N. Silva, J. J. Keizer, T. Warneke, and J. M. C. Pereira, \u0026ldquo;A promising new approach to estimate drought indices for fire danger assessment using remotely sensed data,\u0026rdquo; Agric For Meteorol, vol. 274, pp. 195\u0026ndash;209, Aug. 2019, doi: 10.1016/j.agrformet.2019.04.015.\u003c/li\u003e\n\u003cli\u003eK. Shojaeizadeh, M. Ahmadi, and A. Dadashi-Roudbari, \u0026ldquo;Investigating drought in Iran using Keetch-Byram drought index (KBDI) and evaluating it with surface soil moisture (SSM) of active microwave sensors,\u0026rdquo; Iranian Journal of Geophysics, vol. 17, no. 2, p. 240, Jul. 2023, doi: 10.30499/IJG.2023.391095.1509.\u003c/li\u003e\n\u003cli\u003eG. Ghazaryan, O. Dubovyk, V. Graw, N. Kussul, and J. Schellberg, \u0026ldquo;Local-scale agricultural drought monitoring with satellite-based multi-sensor time-series,\u0026rdquo; GIsci Remote Sens, vol. 57, no. 5, pp. 704\u0026ndash;718, Jul. 2020, doi: 10.1080/15481603.2020.1778332.\u003c/li\u003e\n\u003cli\u003eB. Su et al., \u0026ldquo;Drought losses in China might double between the 1.5 \u0026deg;C and 2.0 \u0026deg;C warming,\u0026rdquo; Proceedings of the National Academy of Sciences, vol. 115, no. 42, pp. 10600\u0026ndash;10605, Oct. 2018, doi: 10.1073/pnas.1802129115.\u003c/li\u003e\n\u003cli\u003eJ. Qiu, Z. Shen, and H. Xie, \u0026ldquo;Drought impacts on hydrology and water quality under climate change,\u0026rdquo; Science of The Total Environment, vol. 858, p. 159854, 2023, doi: https://doi.org/10.1016/j.scitotenv.2022.159854.\u003c/li\u003e\n\u003cli\u003eE. Li, J. Zhao, W. Zhang, and X. Yang, \u0026ldquo;Spatial-temporal patterns of high-temperature and drought during the maize growing season under current and future climate changes in northeast China,\u0026rdquo; J Sci Food Agric, vol. 103, no. 12, pp. 5709\u0026ndash;5716, Sep. 2023, doi: 10.1002/jsfa.12650.\u003c/li\u003e\n\u003cli\u003eF. Zhao and Y. Liu, \u0026ldquo;Important meteorological predictors for long-range wildfires in China,\u0026rdquo; For Ecol Manage, vol. 499, p. 119638, 2021, doi: https://doi.org/10.1016/j.foreco.2021.119638.\u003c/li\u003e\n\u003cli\u003eH. Jia, F. Chen, E. Du, and L. Wang, \u0026ldquo;Drought Vulnerability Curves Based on Remote Sensing and Historical Disaster Dataset,\u0026rdquo; Remote Sens (Basel), vol. 15, no. 3, 2023, doi: 10.3390/rs15030858.\u003c/li\u003e\n\u003cli\u003eC. A. Alonzo, J. M. Galabay, M. N. Macatangay, M. B. Magpayo, and R. Ramirez, \u0026ldquo;Drought Risk Assessment and Monitoring of Ilocos Norte Province in the Philippines Using Satellite Remote Sensing and Meteorological Data,\u0026rdquo; AgriEngineering, vol. 5, no. 2, pp. 720\u0026ndash;739, 2023, doi: 10.3390/agriengineering5020045.\u003c/li\u003e\n\u003cli\u003eJ. Li et al., \u0026ldquo;Robust Meteorological Drought Prediction Using Antecedent SST Fluctuations and Machine Learning,\u0026rdquo; Water Resour Res, vol. 57, no. 8, Aug. 2021, doi: 10.1029/2020WR029413.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Drought, Machine Learning, EVI, SPI, SPEI, KBDI, ANN","lastPublishedDoi":"10.21203/rs.3.rs-5259358/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5259358/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTimely and accurate monitoring of the beginning and development of drought in China is significant in decreasing losses from drought. The present study contributes to a comprehensive spatio-temporal analysis of drought over the Ningxia Hui (northwestern China) from 2003\u0026ndash;2023. We determined the moisture content and vegetation using MODIS satellite data. The Enhanced Vegetation Index (EVI), the Land Surface Temperature (LST), the Standardized Precipitation Index (SPI-1, SPI-3, SPI-6, SPI-9 and SPI-12), and the Standardized Precipitation-Evapotranspiration Index (SPEI-1, SPEI-3, SPEI-6, SPEI-9, and SPEI-12), were calculated. SPEI at 1\u0026ndash;12 months timescales and the Keetch-Byram Drought Index (KBDI) were adopted to characterize drought events over the Ningxia region from 2003 to 2023. Future drought predictions were determined based on SPI at 1\u0026ndash;12 months timescales using an artificial neural network (ANN) and cellular automata (CA) machine learning approaches. The CA-ANN model was used to validate drought prediction. The results showed: (1) the EVI declined from 0.38 to 0.33 from 2003\u0026ndash;2023. This declining EVI indicates that the annual average of vegetation was decreased ; (2) The KBDI increased from 581.33 in 2003 to 681.091 in 2023, reflecting aggrading aridity with the soil moisture drying out; (3) SPI decreased from 0.7 in 2003 to -1.835 in 2023 and the SPEI varied from 0.5 to \u0026minus;\u0026thinsp;1.898 in the same period, (4) SPEI results in 2003 highlight western and southern parts highly affected by drought; (6) drought prediction from CA-ANN display that the SPI and SPEI expected in 2033 will further decrease and can cause more frequent drought. The study concluded that the ever-declining drought conditions in the Ningxia region over the past two decades have manifested drastic changes in the drought conditions.\u003c/p\u003e","manuscriptTitle":"Spatio-temporal Monitoring of Drought using Machine Learning approach and Remote Sensing Techniques in Ningxia","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-10-16 15:42:48","doi":"10.21203/rs.3.rs-5259358/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5f7ee89b-7584-4f74-9419-8f3517b8704d","owner":[],"postedDate":"October 16th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-11-14T21:53:15+00:00","versionOfRecord":[],"versionCreatedAt":"2024-10-16 15:42:48","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5259358","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5259358","identity":"rs-5259358","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}