An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil

An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil | 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 An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil Rodrigo Soares Vieira dos Santos, Michel Macedo Meira This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6710541/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 This study presents an automated methodology for generating seasonal mosaics from Landsat 8 imagery using NDVI time series within the Google Earth Engine (GEE) platform. The method identifies the driest and wettest months of the year based on average NDVI values, allowing for the selection of an optimal temporal window to filter image collections and generate mosaics. The approach was applied in a semi-arid region in the western portion of Bahia, Brazil, which was the target of a geological mapping project. The region’s well-defined climatic variation favored the application of the technique. Preprocessing steps included spatial, temporal, and cloud cover filtering, which reduced the dataset to 444 images. NDVI was calculated for each image, and monthly averages were used to define the driest months (September, October, and November) and wettest months (January, February, and March). Based on this information, separate mosaics were generated for each period. The results showed clear differences between mosaics, both visually and in terms of mean NDVI values (0.16 for the dry season and 0.34 for the wet season). The dry season mosaic, characterized mainly by exposed soil, was selected for use in geological mapping. By automating steps such as optimal temporal window identification and image selection, the method significantly reduced processing time compared to manual approaches. The methodology can be easily adapted for use in other regions and applications, especially in areas with marked climatic seasonality. Remote sensing Google Earth Engine Geological mapping Cloud-based geospatial processing Spectral indices Figures Figure 1 Figure 2 Figure 3 1. Introduction Optical remote sensing imagery acquired by different types of sensors provides highly useful information for the characterization and analysis of the spatial distribution of various surface materials, such as rocks, soils, vegetation, and water bodies. These data are widely used across multiple fields of knowledge, including land use and land cover studies, environmental monitoring, habitat mapping, geological cartography, and territorial management. Materials absorb or reflect varying amounts of energy across the electromagnetic spectrum depending on their composition. The most commonly used spectral regions in remote sensing applications are the visible (VIS), near-infrared (NIR), shortwave infrared (SWIR), and thermal infrared (TIR) (Chaves and Rocha, 2006 ). Images from the Landsat 8 mission sensors are extensively used in a variety of applications due to their global coverage, broad temporal resolution, and free accessibility, which facilitates large-scale use. However, depending on the user’s objective, it may be necessary to restrict the analysis to specific times of the year, such as the dry or wet season. Studies involving vegetation cover and land use, for example, may benefit from images acquired during wetter months, while soil analysis, geological mapping and mineral exploration tend to benefit from data collected during drier periods. The manual selection of appropriate images, combined with the already complex workflow of remote sensing data acquisition and processing, can represent a significant time investment within a project schedule. Given this context, this study proposes an automated and reproducible methodology (Fig. 1 ), developed using the Google Earth Engine (GEE) platform, for generating seasonal mosaics from NDVI (Normalized Difference Vegetation Index) (Rouse et al., 1974 ) time series derived from Landsat 8 data. The method enables the objective identification of the driest and wettest periods of the year and the generation of representative mosaics for each season, optimizing processing time and addressing the specific needs of various applications. 2. Materials and Methods 2.1. Study Area The study area corresponds to the Bom Jesus da Lapa topographic sheet, located in the state of Bahia, Brazil. It covers approximately 3,000 km², between the coordinates 13°00′S to 13°30′S and 43°00′W to 43°30′W. The region is situated in the western portion of Bahia and partially or fully includes the municipalities of Bom Jesus da Lapa, Riacho de Santana, Paratinga, Serra do Ramalho, and Sítio do Mato. The prevailing climate is semi-arid, with a well-defined dry season. Annual average temperatures range from 24°C to 28°C, and annual precipitation typically ranges between 600 mm and 800 mm. This area was selected for the study because it was the target of a 1:100,000-scale geological mapping project (Meira, 2024 ), which involved the use of spectral indices and band combinations derived from Landsat 8 imagery. As this type of application benefits from dry-season scenes, and the region presents a marked climatic seasonality, the opportunity arose to develop a methodology for automatically detecting the ideal temporal window for image selection and generating a cloud-free mosaic that covers the entire study area. 2.2. Landsat 8 Collection and Preprocessing To enable the implementation of the methodology (Fig. 1 ) and ensure efficient processing, Google Earth Engine (GEE) was selected as the processing platform. GEE is a cloud-based platform for geospatial data analysis and processing at global scale, which leverages Google’s computational infrastructure to support applications in various scientific fields. It provides an extensive catalog of remote sensing datasets at different processing levels, ready to be used in advanced analyses. Access to the platform is done directly via a web browser, and users are provided with both a JavaScript API and an integrated development environment (IDE), enabling fast and efficient script execution and visualization of results—without the need to install any local software (Gorelick et al., 2017 ). Numerous studies around the world have utilized the power of GEE in a variety of applications, such as species distribution modeling (Dong et al., 2016 ), land use and land cover classification (Medeiro, 2018 ; Oliveira, 2016 ; Souza et al., 2020 ; Wang et al., 2020 ), urban growth analysis (Patel et al., 2015 ), and geomorphological mapping (Boothroyd et al., 2021 ), among others. The image collection selected for this study was the Landsat 8 Surface Reflectance, Collection 2, available on the Google Earth Engine platform. This dataset contains scenes acquired by the Landsat 8 OLI/TIRS sensors, corrected to surface reflectance and provided at a 30-meter spatial resolution since 2013. Although extensive and comprehensive, the database includes a large number of scenes unrelated to the area of interest, which could affect processing efficiency. To reduce the data volume and retain only the most suitable images for the study, several filters were applied to the collection. This preprocessing step improved performance and ensured that the selected scenes aligned with the study’s spatial and temporal requirements. Each scene contains three visible bands, one near-infrared (VNIR) band, two shortwave infrared (SWIR) bands, and one thermal infrared (TIR) band, as well as quality assessment (QA) bands. First, a spatial filter was applied to retain only scenes containing pixels within the study area. This was accomplished using the filterBounds function with a vector file defining the project boundaries. Next, a temporal filter was used to restrict the collection to the desired acquisition period, applying the filterDate function. We selected images acquired between January 1, 2014, and December 31, 2024, which provided a sufficiently long time series for analyzing NDVI behavior in the region. Finally, a cloud cover filter was applied. Each scene includes metadata indicating the percentage of pixels affected by clouds. Based on this information, we used the filterMetadata function to retain only images with cloud cover below a defined threshold. After applying all filters, the resulting dataset included 444 scenes, which were used for the NDVI analysis. 2.3. NDVI Calculation To continue the methodology, the Normalized Difference Vegetation Index (NDVI) was calculated for each image in the filtered collection using the following equation: NDVI = (NIR - Red) / (NIR + Red) In this case, the NIR and Red bands correspond to SR_B5 and SR_B4 of the Landsat 8 Surface Reflectance product, respectively. To verify that the processing was working correctly, a time series chart was generated (Fig. 2 ), allowing the visualization of NDVI variation over the study area throughout the analyzed period. At this stage, it is already possible to visually interpret the driest season in order to manually select the most suitable images for geological mapping purposes. However, the process can be further optimized by implementing an automated method to select the ideal temporal window for mosaic generation. 2.4. Monthly NDVI Statistics and Season Identification To enable the automatic identification of dry and wet seasons, the next step consisted of performing a statistical analysis of the monthly mean NDVI values across the entire time series. This approach makes it possible to determine the driest months (with lower NDVI values) and the wettest months (with higher NDVI values). Based on the metadata of each image, the month of acquisition was extracted. Then, the mean NDVI was calculated for each calendar month using the ee.Reducer.mean statistical reducer. This process allowed the identification of the driest and wettest months in the dataset, which were then used to generate the seasonal mosaics of interest. 2.5. Seasonal Mosaic Generation To generate the dry and wet season mosaics, the three months with the lowest and highest mean NDVI values, respectively, were selected. For the study area, the driest months were September, October, and November, while the wettest months were January, February, and March. The image collection was then filtered to include only scenes acquired during the selected months. Two separate mosaics were subsequently generated—one for the dry season and one for the wet season—to visualize the results of the analysis (Fig. 3 ). The Google Earth Engine script used in this study is available in the data availability section of this article. 3. Results and Discussions The chart in Fig. 2 shows a clear annual NDVI variation pattern, which is consistent with the expected behavior for a region with a semi-arid climate. This pattern confirms the presence of distinct dry and wet seasons. The interpretation of the time series aligns well with the months automatically identified by the methodology—August, September, and October as the driest, and January, February, and March as the wettest—which were used for mosaic generation. The resulting seasonal mosaics (Fig. 3 ) display visually distinct characteristics. The wet season mosaic reveals a dense and homogeneous vegetative cover, which corresponds to a higher mean NDVI value of 0.34. In contrast, the dry season mosaic exhibits reduced vegetation cover and a predominance of exposed soil, reflected in a lower mean NDVI value of 0.16. This dry season mosaic was selected for use in the geological mapping of the study area due to its enhanced exposure of surface features. Using Google Earth Engine as the processing platform proved to be an effective choice, allowing for the rapid and automated execution of several tasks that would otherwise be time-consuming if done manually—such as cloud-free image selection and optimal temporal filtering. It is important to note that the effectiveness of the proposed methodology may be influenced by the climatic characteristics of the study area. In regions where dry and wet seasons are less pronounced, a more careful evaluation may be necessary to select an appropriate temporal window and generate a suitable mosaic for the intended application. 4. Conclusion This work presents an automated method for selecting temporal intervals based on NDVI and generating seasonal mosaics. The methodology proved capable of correctly distinguishing between dry and wet periods, as evidenced by both the difference in the mean NDVI values of the resulting mosaics and their distinct visual characteristics. The application of this method resulted in a substantial reduction in the time and effort required for the task compared to traditional approaches for image selection and mosaic generation. The method is adaptable and can be applied in other geographic contexts. The automated workflow, based on NDVI time series from freely available satellite data and implemented in Google Earth Engine, ensures scalability and reproducibility, making it suitable for large-scale applications worldwide. To function effectively, the method depends on the presence of sufficient climatic seasonality that can be detected through NDVI variation. It may also be important to test the methodology using other spectral indices that are potentially more sensitive to specific regional climatic conditions. Declarations Ethical Responsibilities of Authors All authors have read, understood, and complied with the journal’s ethical standards as outlined in the "Ethical responsibilities of authors" section in the Instructions for Authors. CRediT authorship contribution statement Rodrigo Soares Vieira dos Santos : Conceptualization, Data curation, Formal analysis, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review & editing. Michel Macedo Meira: Validation, Formal analysis. Declaration of generative AI and AI-assisted technologies in the writing process During the preparation of this work the author used ChatGPT and Grammarly in order to improve language and readability. After using these tools/services, the author reviewed and edited the content as needed and takes full responsibility for the content of the publication. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Funding declaration The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Acknowledgements The author(s) would like to thank the Geological Survey of Brazil – SGB/CPRM and the program PPGM-UEFS, as well as colleagues, supervisors, and professors for providing the structure and an environment conducive to knowledge development, which made this work possible. Data Availability The Google Earth Engine script developed for this study, which includes the full processing workflow for seasonal NDVI-based mosaic generation, is publicly available at the following link: https://code.earthengine.google.com/f522d9008e37877669ebb9bcaffd4af8. All satellite imagery used in the study is available from the Google Earth Engine Data Catalog, specifically from the Landsat 8 Surface Reflectance Collection 2 Tier 1 dataset. References Boothroyd, R.J., Williams, R.D., Hoey, T.B., Barrett, B., Prasojo, O.A., 2021. Applications of Google Earth Engine in fluvial geomorphology for detecting river channel change. WIREs Water 8, e21496. https://doi.org/10.1002/wat2.1496 Chaves, J., Rocha, W., 2006. Geotecnologias trilhando novos caminhos nas geociências. Dong, J., Xiao, X., Menarguez, M.A., Zhang, G., Qin, Y., Thau, D., Biradar, C., Moore, B., 2016. Mapping paddy rice planting area in northeastern Asia with Landsat 8 images, phenology-based algorithm and Google Earth Engine. Remote Sens. Environ., Landsat 8 Science Results 185, 142–154. https://doi.org/10.1016/j.rse.2016.02.016 Gorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., Moore, R., 2017. Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sens. Environ., Big Remotely Sensed Data: tools, applications and experiences 202, 18–27. https://doi.org/10.1016/j.rse.2017.06.031 Medeiro, F.S.L., 2018. Classificação Do Uso E Cobertura Das Terras E Quantificação Dos Padrões Espaciais De Áreas De Caatinga Em Séries Temporais. Universidade Estadual de Feira de Santana. Meira, M.M., 2024. Geologia e recursos minerais da folha Bom Jesus da Lapa - SD.23-X-D-I. Oliveira, L.P., 2016. Uso e cobertura das terras no entorno da Baía de Todos os Santos, Bahia, Brasil: mudanças dos padrões espaciais em séries temporais. Universidade Estadual de Feira de Santana. Patel, N.N., Angiuli, E., Gamba, P., Gaughan, A., Lisini, G., Stevens, F.R., Tatem, A.J., Trianni, G., 2015. Multitemporal settlement and population mapping from Landsat using Google Earth Engine. Int. J. Appl. Earth Obs. Geoinformation 35, 199–208. https://doi.org/10.1016/j.jag.2014.09.005 Rouse, J.W., Jr., Haas, R.H., Schell, J.A., Deering, D.W., 1974. Monitoring Vegetation Systems in the Great Plains with Erts, in: NASA Special Publication. p. 309. Souza, C.M., Z. Shimbo, J., Rosa, M.R., Parente, L.L., A. Alencar, A., Rudorff, B.F.T., Hasenack, H., Matsumoto, M., G. Ferreira, L., Souza-Filho, P.W.M., de Oliveira, S.W., Rocha, W.F., Fonseca, A.V., Marques, C.B., Diniz, C.G., Costa, D., Monteiro, D., Rosa, E.R., Vélez-Martin, E., Weber, E.J., Lenti, F.E.B., Paternost, F.F., Pareyn, F.G.C., Siqueira, J.V., Viera, J.L., Neto, L.C.F., Saraiva, M.M., Sales, M.H., Salgado, M.P.G., Vasconcelos, R., Galano, S., Mesquita, V.V., Azevedo, T., 2020. Reconstructing Three Decades of Land Use and Land Cover Changes in Brazilian Biomes with Landsat Archive and Earth Engine. Remote Sens. 12, 2735. https://doi.org/10.3390/rs12172735 Wang, L., Diao, C., Xian, G., Yin, D., Lu, Y., Zou, S., Erickson, T.A., 2020. A summary of the special issue on remote sensing of land change science with Google earth engine. Remote Sens. Environ. 248, 112002. https://doi.org/10.1016/j.rse.2020.112002 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-6710541","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":476993215,"identity":"4614cca4-61b3-43bf-9cee-242b262dbacb","order_by":0,"name":"Rodrigo Soares Vieira dos Santos","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABDUlEQVRIie3RsUrEMBjA8S8E4hLoGqHYV4jccBwU+yAuCUInuaVLB4cchbgcN1ccfAVFcE4I1KXoA5yDcItjb5FDjkNUODpcvY6C+S3JF/KHQAA872/iIL4WDGAAYgD2cyz6JmnPpMXtT4aXxUPz+hEDf8TW5RfPSXSq8NtSw3iodidhXWWlnKXAHRG2ruby7sWQ0ZWGLDS7E8bOByCnDg4Lyq0ic3FcCjKgNciy42GtJGis2jwl/RKxchBgCnaiDbphAi9o/ktCqwykSmmACbeT2Zm8Da1GZc4y1pUcFPdotY6PSOAWS/V+kkTXhWsaHo+7km9IA90O3CDNtr/Tad3aRwpws+e+53ne//IJso1ZcsuTiBwAAAAASUVORK5CYII=","orcid":"","institution":"Geological Survey of Brazil","correspondingAuthor":true,"prefix":"","firstName":"Rodrigo","middleName":"Soares Vieira dos","lastName":"Santos","suffix":""},{"id":476993216,"identity":"746bbb41-0250-4174-8022-10aee27858ce","order_by":1,"name":"Michel Macedo Meira","email":"","orcid":"","institution":"Geological Survey of Brazil","correspondingAuthor":false,"prefix":"","firstName":"Michel","middleName":"Macedo","lastName":"Meira","suffix":""}],"badges":[],"createdAt":"2025-05-20 19:38:06","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6710541/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6710541/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":85649772,"identity":"e1f3f150-3393-4510-82e2-3f1b85e26744","added_by":"auto","created_at":"2025-06-30 09:03:17","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":918924,"visible":true,"origin":"","legend":"\u003cp\u003eMethodological flowchart illustrating the main steps performed for generating seasonal NDVI-based mosaics using Landsat 8 imagery in Google Earth Engine.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6710541/v1/2c2fbe531c9825019951eaa0.png"},{"id":85649770,"identity":"90f85b08-7265-4ee8-a9a4-0fb5a15f529b","added_by":"auto","created_at":"2025-06-30 09:03:17","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":340939,"visible":true,"origin":"","legend":"\u003cp\u003eNDVI time series chart derived from Landsat 8 imagery over the study area between 2014 and 2024. The chart reveals recurring seasonal patterns, with lower NDVI values during the dry season and higher values during the wet season.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6710541/v1/9deca2d9019563feb763c00d.jpeg"},{"id":85649224,"identity":"af024817-10ea-4618-acc3-f24bc04bef21","added_by":"auto","created_at":"2025-06-30 08:55:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":3781930,"visible":true,"origin":"","legend":"\u003cp\u003eSeasonal mosaics generated for the study area. (A) Mosaic for the dry season (September–November), showing predominance of exposed soil and low vegetation cover. (B) Mosaic for the wet season (January–March), characterized by dense and homogeneous vegetation.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6710541/v1/7ce7a6e2c33355b7fb4fd412.png"},{"id":88020322,"identity":"a1df3e52-de9b-4829-833b-2fb214a9a660","added_by":"auto","created_at":"2025-07-31 13:47:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5935049,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6710541/v1/20e0ac47-4f4b-48ae-8cf7-0f1538bff98c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eOptical remote sensing imagery acquired by different types of sensors provides highly useful information for the characterization and analysis of the spatial distribution of various surface materials, such as rocks, soils, vegetation, and water bodies. These data are widely used across multiple fields of knowledge, including land use and land cover studies, environmental monitoring, habitat mapping, geological cartography, and territorial management.\u003c/p\u003e \u003cp\u003eMaterials absorb or reflect varying amounts of energy across the electromagnetic spectrum depending on their composition. The most commonly used spectral regions in remote sensing applications are the visible (VIS), near-infrared (NIR), shortwave infrared (SWIR), and thermal infrared (TIR) (Chaves and Rocha, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2006\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eImages from the Landsat 8 mission sensors are extensively used in a variety of applications due to their global coverage, broad temporal resolution, and free accessibility, which facilitates large-scale use. However, depending on the user\u0026rsquo;s objective, it may be necessary to restrict the analysis to specific times of the year, such as the dry or wet season. Studies involving vegetation cover and land use, for example, may benefit from images acquired during wetter months, while soil analysis, geological mapping and mineral exploration tend to benefit from data collected during drier periods. The manual selection of appropriate images, combined with the already complex workflow of remote sensing data acquisition and processing, can represent a significant time investment within a project schedule.\u003c/p\u003e \u003cp\u003eGiven this context, this study proposes an automated and reproducible methodology (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), developed using the Google Earth Engine (GEE) platform, for generating seasonal mosaics from NDVI (Normalized Difference Vegetation Index) (Rouse et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e1974\u003c/span\u003e) time series derived from Landsat 8 data. The method enables the objective identification of the driest and wettest periods of the year and the generation of representative mosaics for each season, optimizing processing time and addressing the specific needs of various applications.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"2. Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Study Area\u003c/h2\u003e \u003cp\u003eThe study area corresponds to the \u003cem\u003eBom Jesus da Lapa\u003c/em\u003e topographic sheet, located in the state of Bahia, Brazil. It covers approximately 3,000 km\u0026sup2;, between the coordinates 13\u0026deg;00\u0026prime;S to 13\u0026deg;30\u0026prime;S and 43\u0026deg;00\u0026prime;W to 43\u0026deg;30\u0026prime;W. The region is situated in the western portion of Bahia and partially or fully includes the municipalities of Bom Jesus da Lapa, Riacho de Santana, Paratinga, Serra do Ramalho, and S\u0026iacute;tio do Mato.\u003c/p\u003e \u003cp\u003eThe prevailing climate is semi-arid, with a well-defined dry season. Annual average temperatures range from 24\u0026deg;C to 28\u0026deg;C, and annual precipitation typically ranges between 600 mm and 800 mm.\u003c/p\u003e \u003cp\u003eThis area was selected for the study because it was the target of a 1:100,000-scale geological mapping project (Meira, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2024\u003c/span\u003e), which involved the use of spectral indices and band combinations derived from Landsat 8 imagery. As this type of application benefits from dry-season scenes, and the region presents a marked climatic seasonality, the opportunity arose to develop a methodology for automatically detecting the ideal temporal window for image selection and generating a cloud-free mosaic that covers the entire study area.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Landsat 8 Collection and Preprocessing\u003c/h2\u003e \u003cp\u003eTo enable the implementation of the methodology (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and ensure efficient processing, Google Earth Engine (GEE) was selected as the processing platform. GEE is a cloud-based platform for geospatial data analysis and processing at global scale, which leverages Google\u0026rsquo;s computational infrastructure to support applications in various scientific fields. It provides an extensive catalog of remote sensing datasets at different processing levels, ready to be used in advanced analyses.\u003c/p\u003e \u003cp\u003eAccess to the platform is done directly via a web browser, and users are provided with both a JavaScript API and an integrated development environment (IDE), enabling fast and efficient script execution and visualization of results\u0026mdash;without the need to install any local software (Gorelick et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eNumerous studies around the world have utilized the power of GEE in a variety of applications, such as species distribution modeling (Dong et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), land use and land cover classification (Medeiro, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Oliveira, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Souza et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2020\u003c/span\u003e), urban growth analysis (Patel et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), and geomorphological mapping (Boothroyd et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), among others.\u003c/p\u003e \u003cp\u003eThe image collection selected for this study was the Landsat 8 Surface Reflectance, Collection 2, available on the Google Earth Engine platform. This dataset contains scenes acquired by the Landsat 8 OLI/TIRS sensors, corrected to surface reflectance and provided at a 30-meter spatial resolution since 2013. Although extensive and comprehensive, the database includes a large number of scenes unrelated to the area of interest, which could affect processing efficiency.\u003c/p\u003e \u003cp\u003eTo reduce the data volume and retain only the most suitable images for the study, several filters were applied to the collection. This preprocessing step improved performance and ensured that the selected scenes aligned with the study\u0026rsquo;s spatial and temporal requirements. Each scene contains three visible bands, one near-infrared (VNIR) band, two shortwave infrared (SWIR) bands, and one thermal infrared (TIR) band, as well as quality assessment (QA) bands.\u003c/p\u003e \u003cp\u003eFirst, a spatial filter was applied to retain only scenes containing pixels within the study area. This was accomplished using the filterBounds function with a vector file defining the project boundaries.\u003c/p\u003e \u003cp\u003eNext, a temporal filter was used to restrict the collection to the desired acquisition period, applying the filterDate function. We selected images acquired between January 1, 2014, and December 31, 2024, which provided a sufficiently long time series for analyzing NDVI behavior in the region.\u003c/p\u003e \u003cp\u003eFinally, a cloud cover filter was applied. Each scene includes metadata indicating the percentage of pixels affected by clouds. Based on this information, we used the filterMetadata function to retain only images with cloud cover below a defined threshold. After applying all filters, the resulting dataset included 444 scenes, which were used for the NDVI analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. NDVI Calculation\u003c/h2\u003e \u003cp\u003eTo continue the methodology, the Normalized Difference Vegetation Index (NDVI) was calculated for each image in the filtered collection using the following equation:\u003c/p\u003e \u003cp\u003eNDVI = (NIR - Red) / (NIR\u0026thinsp;+\u0026thinsp;Red)\u003c/p\u003e \u003cp\u003eIn this case, the NIR and Red bands correspond to SR_B5 and SR_B4 of the Landsat 8 Surface Reflectance product, respectively.\u003c/p\u003e \u003cp\u003eTo verify that the processing was working correctly, a time series chart was generated (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e), allowing the visualization of NDVI variation over the study area throughout the analyzed period. At this stage, it is already possible to visually interpret the driest season in order to manually select the most suitable images for geological mapping purposes. However, the process can be further optimized by implementing an automated method to select the ideal temporal window for mosaic generation.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Monthly NDVI Statistics and Season Identification\u003c/h2\u003e \u003cp\u003eTo enable the automatic identification of dry and wet seasons, the next step consisted of performing a statistical analysis of the monthly mean NDVI values across the entire time series. This approach makes it possible to determine the driest months (with lower NDVI values) and the wettest months (with higher NDVI values).\u003c/p\u003e \u003cp\u003eBased on the metadata of each image, the month of acquisition was extracted. Then, the mean NDVI was calculated for each calendar month using the \u003cem\u003eee.Reducer.mean\u003c/em\u003e statistical reducer. This process allowed the identification of the driest and wettest months in the dataset, which were then used to generate the seasonal mosaics of interest.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Seasonal Mosaic Generation\u003c/h2\u003e \u003cp\u003eTo generate the dry and wet season mosaics, the three months with the lowest and highest mean NDVI values, respectively, were selected. For the study area, the driest months were September, October, and November, while the wettest months were January, February, and March.\u003c/p\u003e \u003cp\u003eThe image collection was then filtered to include only scenes acquired during the selected months. Two separate mosaics were subsequently generated\u0026mdash;one for the dry season and one for the wet season\u0026mdash;to visualize the results of the analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe Google Earth Engine script used in this study is available in the data availability section of this article.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results and Discussions","content":"\u003cp\u003eThe chart in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows a clear annual NDVI variation pattern, which is consistent with the expected behavior for a region with a semi-arid climate. This pattern confirms the presence of distinct dry and wet seasons. The interpretation of the time series aligns well with the months automatically identified by the methodology\u0026mdash;August, September, and October as the driest, and January, February, and March as the wettest\u0026mdash;which were used for mosaic generation.\u003c/p\u003e \u003cp\u003eThe resulting seasonal mosaics (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) display visually distinct characteristics. The wet season mosaic reveals a dense and homogeneous vegetative cover, which corresponds to a higher mean NDVI value of 0.34. In contrast, the dry season mosaic exhibits reduced vegetation cover and a predominance of exposed soil, reflected in a lower mean NDVI value of 0.16. This dry season mosaic was selected for use in the geological mapping of the study area due to its enhanced exposure of surface features.\u003c/p\u003e \u003cp\u003eUsing Google Earth Engine as the processing platform proved to be an effective choice, allowing for the rapid and automated execution of several tasks that would otherwise be time-consuming if done manually\u0026mdash;such as cloud-free image selection and optimal temporal filtering.\u003c/p\u003e \u003cp\u003eIt is important to note that the effectiveness of the proposed methodology may be influenced by the climatic characteristics of the study area. In regions where dry and wet seasons are less pronounced, a more careful evaluation may be necessary to select an appropriate temporal window and generate a suitable mosaic for the intended application.\u003c/p\u003e"},{"header":"4. Conclusion","content":"\u003cp\u003eThis work presents an automated method for selecting temporal intervals based on NDVI and generating seasonal mosaics. The methodology proved capable of correctly distinguishing between dry and wet periods, as evidenced by both the difference in the mean NDVI values of the resulting mosaics and their distinct visual characteristics.\u003c/p\u003e \u003cp\u003eThe application of this method resulted in a substantial reduction in the time and effort required for the task compared to traditional approaches for image selection and mosaic generation. The method is adaptable and can be applied in other geographic contexts. The automated workflow, based on NDVI time series from freely available satellite data and implemented in Google Earth Engine, ensures scalability and reproducibility, making it suitable for large-scale applications worldwide.\u003c/p\u003e \u003cp\u003eTo function effectively, the method depends on the presence of sufficient climatic seasonality that can be detected through NDVI variation. It may also be important to test the methodology using other spectral indices that are potentially more sensitive to specific regional climatic conditions.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthical Responsibilities of Authors\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll authors have read, understood, and complied with the journal’s ethical standards as outlined in the \"Ethical responsibilities of authors\" section in the Instructions for Authors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCRediT authorship contribution statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRodrigo Soares Vieira dos Santos\u003c/strong\u003e: Conceptualization, Data curation, Formal analysis, Methodology, Software, Validation, Visualization, Writing – original draft, Writing – review \u0026amp; editing. \u003cstrong\u003eMichel Macedo Meira: \u003c/strong\u003eValidation, Formal analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of generative AI and AI-assisted technologies in the writing process\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDuring the preparation of this work the author used ChatGPT and Grammarly in order to improve language and readability. After using these tools/services, the author reviewed and edited the content as needed and takes full responsibility for the content of the publication.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of competing interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that no funds, grants, or other support were received during the preparation of this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author(s) would like to thank the Geological Survey of Brazil – SGB/CPRM and the program PPGM-UEFS, as well as colleagues, supervisors, and professors for providing the structure and an environment conducive to knowledge development, which made this work possible.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Google Earth Engine script developed for this study, which includes the full processing workflow for seasonal NDVI-based mosaic generation, is publicly available at the following link:\u003c/p\u003e\n\u003cp\u003ehttps://code.earthengine.google.com/f522d9008e37877669ebb9bcaffd4af8.\u003c/p\u003e\n\u003cp\u003eAll satellite imagery used in the study is available from the Google Earth Engine Data Catalog, specifically from the Landsat 8 Surface Reflectance Collection 2 Tier 1 dataset.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBoothroyd, R.J., Williams, R.D., Hoey, T.B., Barrett, B., Prasojo, O.A., 2021. Applications of Google Earth Engine in fluvial geomorphology for detecting river channel change. WIREs Water 8, e21496. https://doi.org/10.1002/wat2.1496\u003c/li\u003e\n\u003cli\u003eChaves, J., Rocha, W., 2006. Geotecnologias trilhando novos caminhos nas geoci\u0026ecirc;ncias.\u003c/li\u003e\n\u003cli\u003eDong, J., Xiao, X., Menarguez, M.A., Zhang, G., Qin, Y., Thau, D., Biradar, C., Moore, B., 2016. Mapping paddy rice planting area in northeastern Asia with Landsat 8 images, phenology-based algorithm and Google Earth Engine. Remote Sens. Environ., Landsat 8 Science Results 185, 142\u0026ndash;154. https://doi.org/10.1016/j.rse.2016.02.016\u003c/li\u003e\n\u003cli\u003eGorelick, N., Hancher, M., Dixon, M., Ilyushchenko, S., Thau, D., Moore, R., 2017. Google Earth Engine: Planetary-scale geospatial analysis for everyone. Remote Sens. Environ., Big Remotely Sensed Data: tools, applications and experiences 202, 18\u0026ndash;27. https://doi.org/10.1016/j.rse.2017.06.031\u003c/li\u003e\n\u003cli\u003eMedeiro, F.S.L., 2018. Classifica\u0026ccedil;\u0026atilde;o Do Uso E Cobertura Das Terras E Quantifica\u0026ccedil;\u0026atilde;o Dos Padr\u0026otilde;es Espaciais De \u0026Aacute;reas De Caatinga Em S\u0026eacute;ries Temporais. Universidade Estadual de Feira de Santana.\u003c/li\u003e\n\u003cli\u003eMeira, M.M., 2024. Geologia e recursos minerais da folha Bom Jesus da Lapa - SD.23-X-D-I.\u003c/li\u003e\n\u003cli\u003eOliveira, L.P., 2016. Uso e cobertura das terras no entorno da Ba\u0026iacute;a de Todos os Santos, Bahia, Brasil: mudan\u0026ccedil;as dos padr\u0026otilde;es espaciais em s\u0026eacute;ries temporais. Universidade Estadual de Feira de Santana.\u003c/li\u003e\n\u003cli\u003ePatel, N.N., Angiuli, E., Gamba, P., Gaughan, A., Lisini, G., Stevens, F.R., Tatem, A.J., Trianni, G., 2015. Multitemporal settlement and population mapping from Landsat using Google Earth Engine. Int. J. Appl. Earth Obs. Geoinformation 35, 199\u0026ndash;208. https://doi.org/10.1016/j.jag.2014.09.005\u003c/li\u003e\n\u003cli\u003eRouse, J.W., Jr., Haas, R.H., Schell, J.A., Deering, D.W., 1974. Monitoring Vegetation Systems in the Great Plains with Erts, in: NASA Special Publication. p. 309.\u003c/li\u003e\n\u003cli\u003eSouza, C.M., Z. Shimbo, J., Rosa, M.R., Parente, L.L., A. Alencar, A., Rudorff, B.F.T., Hasenack, H., Matsumoto, M., G. Ferreira, L., Souza-Filho, P.W.M., de Oliveira, S.W., Rocha, W.F., Fonseca, A.V., Marques, C.B., Diniz, C.G., Costa, D., Monteiro, D., Rosa, E.R., V\u0026eacute;lez-Martin, E., Weber, E.J., Lenti, F.E.B., Paternost, F.F., Pareyn, F.G.C., Siqueira, J.V., Viera, J.L., Neto, L.C.F., Saraiva, M.M., Sales, M.H., Salgado, M.P.G., Vasconcelos, R., Galano, S., Mesquita, V.V., Azevedo, T., 2020. Reconstructing Three Decades of Land Use and Land Cover Changes in Brazilian Biomes with Landsat Archive and Earth Engine. Remote Sens. 12, 2735. https://doi.org/10.3390/rs12172735\u003c/li\u003e\n\u003cli\u003eWang, L., Diao, C., Xian, G., Yin, D., Lu, Y., Zou, S., Erickson, T.A., 2020. A summary of the special issue on remote sensing of land change science with Google earth engine. Remote Sens. Environ. 248, 112002. https://doi.org/10.1016/j.rse.2020.112002\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":"Remote sensing, Google Earth Engine, Geological mapping, Cloud-based geospatial processing, Spectral indices","lastPublishedDoi":"10.21203/rs.3.rs-6710541/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6710541/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThis study presents an automated methodology for generating seasonal mosaics from Landsat 8 imagery using NDVI time series within the Google Earth Engine (GEE) platform. The method identifies the driest and wettest months of the year based on average NDVI values, allowing for the selection of an optimal temporal window to filter image collections and generate mosaics. The approach was applied in a semi-arid region in the western portion of Bahia, Brazil, which was the target of a geological mapping project. The region\u0026rsquo;s well-defined climatic variation favored the application of the technique.\u003c/p\u003e \u003cp\u003ePreprocessing steps included spatial, temporal, and cloud cover filtering, which reduced the dataset to 444 images. NDVI was calculated for each image, and monthly averages were used to define the driest months (September, October, and November) and wettest months (January, February, and March). Based on this information, separate mosaics were generated for each period. The results showed clear differences between mosaics, both visually and in terms of mean NDVI values (0.16 for the dry season and 0.34 for the wet season). The dry season mosaic, characterized mainly by exposed soil, was selected for use in geological mapping.\u003c/p\u003e \u003cp\u003eBy automating steps such as optimal temporal window identification and image selection, the method significantly reduced processing time compared to manual approaches. The methodology can be easily adapted for use in other regions and applications, especially in areas with marked climatic seasonality.\u003c/p\u003e","manuscriptTitle":"An Automated Approach for Seasonal Mosaic Generation Using NDVI Time Series and Landsat 8 Data: A Case Study in a Semi-Arid Region of Brazil","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-30 08:55:12","doi":"10.21203/rs.3.rs-6710541/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":"579adfa5-b52f-40e4-ae56-10b20ee6d9ae","owner":[],"postedDate":"June 30th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-31T13:38:56+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-30 08:55:12","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6710541","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6710541","identity":"rs-6710541","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}