Remote Sensing Applications-Society and Environment最新文献

{"title":"Satellite-based estimation of net radiation to support evapotranspiration modeling in agriculture","authors":"Chutimon Phoemwong,&nbsp;Rungrat Wattan,&nbsp;Somjet Pattarapanitchai,&nbsp;Serm Janjai","doi":"10.1016/j.rsase.2025.101746","DOIUrl":"10.1016/j.rsase.2025.101746","url":null,"abstract":"<div><div>Net radiation (<em>R</em>_<em>n</em>) is a fundamental variable in the surface energy balance and serves as a key input for estimating evapotranspiration (<em>ET</em>_<em>0</em>), which is critical for agricultural water management and irrigation planning. Accurate estimation of <em>R</em>_<em>n</em> is particularly important in large-scale agricultural regions where ground-based measurements are limited or unavailable. This study aims to investigate the spatiotemporal variation of surface <em>R</em>_<em>n</em> and to develop simplified multiple linear regression models for its estimation using satellite-derived atmospheric variables. The selected input variables chosen for their direct or indirect influence on <em>ET</em>_<em>0</em> include downward shortwave radiation (<em>S</em>_<em>d</em>) and the brightness temperature difference between bands 31 and 32 from MODIS, which indicates atmospheric water vapor content (<em>WP</em>). These are supplemented by relative humidity (<em>RH</em>), air temperature (<em>T</em>_<em>air</em>), and cloud cover (<em>C</em>), obtained from NCEP/NCAR reanalysis data. Ground-based observations of <em>R</em>_<em>n</em> were used as reference data to develop and validate the model. The dataset was divided into two parts: 2017–2021 for model development and 2022–2024 for validation. The resulting linear model showed high accuracy, with an <em>R</em>^<em>2</em> of 0.96, <em>RMSE</em> of 21.6 %, and <em>MBE</em> of −6.4 %. The validated model was applied to produce spatial <em>R</em>_<em>n</em> maps, which demonstrated strong agreement with in-situ data and effectively represented spatial and temporal variation. This modeling approach enhances the ability to estimate <em>R</em>_<em>n</em> over large agricultural areas, thereby supporting more reliable <em>ET</em>_<em>0</em> estimation and water resource management.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101746"},"PeriodicalIF":4.5,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Assessing future flood risk using remote sensing and explainable machine learning: A case study in the Beijing-Tianjin-Hebei region","authors":"Siying Li ,&nbsp;Yi Tang ,&nbsp;Yuting Zhao ,&nbsp;Xiaojun Ning ,&nbsp;Yifan Zhang ,&nbsp;Siran Lv ,&nbsp;Chenshu Liu","doi":"10.1016/j.rsase.2025.101742","DOIUrl":"10.1016/j.rsase.2025.101742","url":null,"abstract":"<div><div>Flood disasters pose increasingly severe threats to densely populated and economically critical regions under changing climate conditions. In this study, we conducted a comprehensive flood risk assessment of the Beijing-Tianjin-Hebei (BTH) region, integrating multi-source remote sensing data and explainable machine learning methods. First, flood inundation areas during the 2023 extreme rainfall event were identified using Sentinel-1 SAR imagery. Based on the assessment framework of hazard, exposure, and vulnerability, key factors influencing flood risk were quantified using a XGBboost model and SHAP (Shapley Additive Explanations) analysis. The results revealed that terrain ruggedness, elevation, precipitation, dependency ratio, and GDP (Gross Domestic Product) were the primary drivers of flood risk distribution. Subsequently, future flood risk patterns for 2030 were projected under four Shared Socioeconomic Pathways (SSP126, SSP245, SSP370, and SSP585), combining projected environmental and socio-economic variables with the established model. The findings indicate a clear trend of flood risk intensification under higher emission scenarios, with high-risk areas expanding significantly under SSP370 and SSP585. These results emphasize the urgent need for differentiated flood management strategies, combining climate mitigation, resilient urban planning, and adaptive infrastructure development to effectively reduce future flood risks. It provides a scientific basis for climate-resilient disaster risk governance in rapidly urbanizing regions.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101742"},"PeriodicalIF":4.5,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222201","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Integrating stepwise residual refinement and explainable AI for interpretable forest volume modeling in Hokkaido, Japan","authors":"Kotaro Iizuka ,&nbsp;Nobuo Ishiyama ,&nbsp;Yasutaka Nakata","doi":"10.1016/j.rsase.2025.101740","DOIUrl":"10.1016/j.rsase.2025.101740","url":null,"abstract":"<div><div>Accurate estimation of forest stem volume is essential for effective forest resource management and carbon accounting. However, spatial heterogeneity in forest conditions often leads to systematic modeling errors, especially across ecological and operational gradients. This study proposes an integrated framework that combines XGBoost-based modeling with a novel Stepwise Residual Refinement (SRR) approach and explainable AI techniques utilizing SHapley Additive exPlanations (SHAP) to enhance both prediction accuracy and model interpretability. The framework was applied to forest inventory and remote sensing data across Hokkaido, Japan, incorporating topographic, climatic, structural, and socioeconomic variables. The initial XGBoost model achieved a root mean square error (RMSE) of 170.21 m³/ha and a percentage RMSE (%RMSE) of 36.90 %. Following the application of SRR corrections, the final model improved significantly, yielding an RMSE of 105.75 m³/ha and a %RMSE of 22.93 %. KernelSHAP analysis revealed region-specific patterns of variable influence, highlighting how environmental and human factors differentially shape forest volume across regions. SHAP-derived zoning delineated clusters of forest quality that aligned with workforce presence and ecological conditions, particularly in conifer-dominated areas. These results demonstrate the importance of integrating explainable AI and spatial refinement to uncover nuanced forest dynamics and support adaptive, data-driven forest management. This study highlights how interpretable machine learning can simultaneously improve predictive accuracy and reveal latent socio-ecological processes that drive forest conditions.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101740"},"PeriodicalIF":4.5,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223402","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Improved soil moisture mapping using an integrated cyclic modeling and bias correction approach","authors":"Yajie Shi ,&nbsp;Wei Dai ,&nbsp;Guangsheng Chen ,&nbsp;Xi Zhang ,&nbsp;Nan Li ,&nbsp;Weijun Fu","doi":"10.1016/j.rsase.2025.101741","DOIUrl":"10.1016/j.rsase.2025.101741","url":null,"abstract":"<div><div>Soil moisture (SM) is crucial for climate change, crop growth estimation, and environmental hazard monitoring. Existing SM products often have low spatial resolution, limiting their use in local-scale studies. While various machine learning (ML) methods have been applied to downscale SM, few studies have explored multiple cyclic modeling or improved downscaling accuracy by recycling qualified stations. In this study, we performed quality control and bias correction on data from the International Soil Moisture Network (ISMN) stations. We obtained qualified sites by cyclic modeling using an extreme gradient boosting (XGBoost) regression model. The predicted bias from cyclic modeling was combined with dynamic environmental variables to correct errors at unqualified sites. Finally, a surface (0–5 cm) soil moisture product with a temporal and spatial resolution of 500 m/day was produced: (1) the XGBoost model described the relationship between SM and environmental variables well, achieving a correlation coefficient (R) of 0.98 and a root mean square error (RMSE) of 0.007 m³/m³ (2) The generated 500 m SM data was comparable to the Soil Moisture Active Passive Level 4 (SMAP-L4) SM data, with 83.2 % of the 1996 points having R &gt; 0.6. The downscaling accuracy is improved by robust cyclic modeling and bias correction techniques, with R, RMSE, and mean absolute error (MAE) improved by 6.5 %, 9.3 %, and 9.6 %, respectively, over single-shot modeling. The estimated results of the surface layer (0–5 cm) soil moisture at 500 m/day can supplement the regional soil moisture database and provide ideas for downscaling soil moisture research.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101741"},"PeriodicalIF":4.5,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223400","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A multi-criteria based optimal niche analysis of seasonal productivity in the Bay of Bengal using MODIS data","authors":"Md Asif Hasan ,&nbsp;Mir Md Tasnim Alam ,&nbsp;Md Zayed Abdur Razzak ,&nbsp;Anika Nawar Mayeesha","doi":"10.1016/j.rsase.2025.101743","DOIUrl":"10.1016/j.rsase.2025.101743","url":null,"abstract":"<div><div>Despite receiving enormous riverine nutrient inputs, the Bay of Bengal (BoB) has a long-standing biogeochemical paradox of comparatively low open-ocean productivity. To understand its long-term trajectory, this study analyzes a consistent two-decade (2003–2022) satellite dataset of Sea Surface Temperature (SST) and Chlorophyll-a (Chl-a) from Moderate Resolution Imaging Spectroradiometer (MODIS) Aqua. Our analysis reveals a significant basin-wide warming trend of +0.019 °C/year and spatially heterogeneous changes in productivity, with a pronounced apparent increasing Chl-a trend in the northern coastal zone (&gt;1 μg/m³/year<strong>)</strong> and a slight decline in open-ocean regions. In areas of nutrient-driven productivity, a weak but statistically significant negative correlation (Pearson's r = −0.204, p &lt; 0.05) between SST and Chl-a anomalies confirms the role of upwelling. As a secondary objective, we developed and applied an exploratory seven-factor Multi-Criteria Evaluation (MCE) model to synthesize these biophysical drivers based on weighted overlay and identify potential productivity hotspots. The model integrates data on SST and Chl-a suitability, thermal and biological fronts, upwelling potential, habitat stability, and depth. The MCE framework successfully identified key productive zones, including persistent biological fronts along the northern coast, characterized by chlorophyll gradients exceeding 6 mg/m³/km, a threshold corresponding to the 90th percentile of all observed gradient values. Notably, with scores &gt;0.7, the spring pre-monsoon period stood out as a time of widespread high productivity, casting doubt on traditional theories of basin-wide oligotrophy. This integrated approach provides a robust quantification of climate-driven trends and offers an exploratory framework for mapping productivity suitability zones, serving as an essential tool for ecosystem-based management in the BoB.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101743"},"PeriodicalIF":4.5,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223398","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Large-scale mangrove mapping in Thailand using multi-sensor ensemble machine learning with sentinel-1/2 and SRTM data","authors":"Surachet Pinkaew ,&nbsp;Werapong Koedsin ,&nbsp;Jonathan Cheung-Wai Chan ,&nbsp;Alfredo Huete","doi":"10.1016/j.rsase.2025.101744","DOIUrl":"10.1016/j.rsase.2025.101744","url":null,"abstract":"<div><div>Mangrove ecosystems provide critical ecological services but face increasing pressure from anthropogenic activities and climate change. Accurate large-scale mapping is essential for effective conservation strategies. We produced a 2024 national mangrove map by merging Sentinel-2 multispectral imagery, Sentinel-1 synthetic-aperture radar (SAR) backscatter and a Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM). The analysis domain comprised all Global Mangrove Watch (2023) polygons with a 2 km buffer. From these layers we derived 23 predictors, including six spectral bands, six vegetation indices (e.g., Normalized Difference Vegetation Index, NDVI; Enhanced Vegetation Index, EVI; Mangrove Vegetation Index, MVI), four radar texture metrics (VV, VH, VV/VH ratio, contrast) and terrain variables(elevation, slope, aspect). Five widely used machine-learning classifiers—Random Forest (RF), Support Vector Machine (SVM), Classification and Regression Tree (CART), Gradient Tree Boosting (GTB) and XGBoost—were combined through soft voting after grid-based hyper-parameter tuning. The ensemble achieved an overall accuracy of 97.0 %, outperforming individual models (95.8–96.9 %). Feature-importance analysis identified MVI as the strongest discriminator (0.209–0.720), followed by VV contrast (0.052–0.097) and elevation (0.044–0.089). The final map shows 2557 km² of mangroves distributed across 24 provinces, with 75 % located along the Andaman Sea coast. By blending complementary optical, radar and topographic information in a fully script-based Google Earth Engine (GEE) workflow, the study delivers an operationally scalable tool for national monitoring that supports conservation planning, carbon accounting and climate-adaptation policies.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101744"},"PeriodicalIF":4.5,"publicationDate":"2025-09-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145223401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Canopy spectral responses of temperate forests to late spring frost and hot drought events assessed with Sentinel-2 NDVI time series","authors":"Michele Dalponte ,&nbsp;Davide Andreatta ,&nbsp;David A. Coomes ,&nbsp;Luca Belelli Marchesini ,&nbsp;Daniele Marinelli ,&nbsp;Loris Vescovo ,&nbsp;Damiano Gianelle","doi":"10.1016/j.rsase.2025.101737","DOIUrl":"10.1016/j.rsase.2025.101737","url":null,"abstract":"<div><div><em>Extreme climatic events (ECEs) are projected to increase due to climate change, but we still have limited understanding of how these events affect the functioning of forest ecosystems. Each species may react differently to ECEs, depending on their ecology, but we lack a regional perspective on these responses. Here we tracked intra-annual changes in the canopy greenness (i.e. NDVI from Sentinel-2 imagery) of 16 tree species growing within</em> 3000 km² <em>of forests of the Italian Alps. The study region was subject to a late spring frost event in May</em> <em>2019, and a hot drought in July 2022, allowing us to quantify species responses to ECEs by comparison of seasonal trends in NDVI observed over the period 2018–2024. The effects of 2019 frost were very localized and mainly affected the canopy spectral response and phenology of</em> Fagus sylvatica L. <em>in areas around 1000 m a.s.l.. There, trees had developed buds and some juvenile leaves when frost occurred, resulting in the wilting or dropping of the earliest leaves, and slower green-up phase but no lasting impacts. The hot drought had its largest impact on</em> Quercus ilex L. <em>forests growing at low elevations: there was a clear decrease in canopy greenness from July onwards in 2022, but no residual impacts were observed the following years. At higher elevations, some species had unusually</em> <em>green</em> <em>canopies in response to the heatwave suggesting they benefitted from warmer conditions</em>.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101737"},"PeriodicalIF":4.5,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222200","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fusion of remote sensing and geochemical data using hybrid Variational Autoencoder- BIRCH deep learning algorithm for copper prospectivity mapping","authors":"Zohre Hoseinzade ,&nbsp;Mobin Saremi ,&nbsp;Mojgan Shojaei ,&nbsp;Ahmad Reza Mokhtari ,&nbsp;Amin Beiranvand Pour ,&nbsp;Seyyed Ataollah Agha Seyyed Mirzabozorg ,&nbsp;Ardeshir Hezarkhani ,&nbsp;Abbas Maghsoudi ,&nbsp;Saeed Yousefi","doi":"10.1016/j.rsase.2025.101738","DOIUrl":"10.1016/j.rsase.2025.101738","url":null,"abstract":"<div><div>Clustering methods are an essential part of machine learning (ML) algorithms and are widely used to integrate a variety of datasets, such as remote sensing, geochemical, and geological data, for mineral prospectivity mapping (MPM). These methods help exploration geologists identify mineralization zones. However, as geospatial datasets become more complex, nonlinear, and high-dimensional, traditional clustering algorithms often fail to handle and analyze them effectively. To address this challenge, this study presents a new unsupervised deep learning (DL) approach called the hybrid Variational Autoencoder- BIRCH (VAE-BIRCH) algorithm, which was applied for porphyry copper prospectivity mapping. The northern sector of Shahr-e-Babak district in southern Iran, which contains numerous porphyry copper deposits, was selected as a case study. ASTER and Landsat 8-OLI satellite remote sensing data were meticulously processed to highlight argillic, silicic, phyllic, propylitic, and iron oxide alteration zones. Factor analysis was applied to stream geochemical data, which demonstrated strong correlation among Copper (Cu), Lead (Pb) and Zinc (Zn). These elements were then used to generate geochemical evidence layers for the study area. These layers were then passed into a VAE, which reduced the data into a lower-dimensional latent space while keeping the important patterns. The VAE created a probability distribution for each sample in the latent space and sampled from it. Then, based on the importance of the input features, the data were passed to the BIRCH clustering algorithm for clustering. The prediction-area (P-A) plot was used to identify anomaly clusters from the background. For comparison, results from the traditional BIRCH algorithm were also generated. The findings showed that the VAE-BIRCH method has a better prediction rate than the BIRCH method. To validate the result of the model, field surveys and laboratory analyses, including microscopic studies and X-ray fluorescence (XRF) analyses, were conducted. These confirmed the presence of minerals associated with porphyry copper mineralization. Based on these results, this paper recommends applying the hybrid VAE-BIRCH algorithm to other copper mineralization provinces and frontier terranes (pristine or remote zones) for mineral exploration targeting worldwide.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101738"},"PeriodicalIF":4.5,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145159870","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Convolutional neural networks for semantic segmentation of aerial images in land cover mapping of environmental protection areas","authors":"Fabricio Bizotto ,&nbsp;Gilson A. Giraldi ,&nbsp;Jose M. Junior ,&nbsp;Victor P.A.V. da Silva ,&nbsp;Ana C.P. Imbelloni ,&nbsp;Mauren L. Andrade ,&nbsp;Jose Marcatto ,&nbsp;Andre Brito","doi":"10.1016/j.rsase.2025.101707","DOIUrl":"10.1016/j.rsase.2025.101707","url":null,"abstract":"<div><div>Environmental Protection Areas (EPAs) in Brazil are key legal instruments for conserving biodiversity, ensuring the sustainable use of natural resources, and supporting the socioeconomic development of local communities. Remote Sensing (RS) has emerged as a practical and cost-efficient alternative for monitoring these regions. In this context, advanced computational techniques - particularly Convolutional Neural Networks (CNNs) — have demonstrated strong performance in the semantic segmentation of RS imagery. This study proposes a low-cost and scalable methodology for land use and land cover mapping in the EPA-Petrópolis region (Rio de Janeiro), based on RGB images from Google Earth and CNN models. This work stands out by offering a low-cost and scalable methodology using RGB imagery from Google Earth, and by introducing the EpaPetroBR dataset–one of the first annotated semantic segmentation datasets focused on the Atlantic Forest biome. The SegNet and U-Net architectures were evaluated across four experimental scenarios. The best overall accuracy (0.87) was obtained in scenario 4, which employed U-Net with the Focal Loss function. Scenario 3, using U-Net with the cross-entropy loss function, achieved comparable accuracy (0.87) and the highest Jaccard Index (IoU) score (0.72). Despite these promising results, some classes — such as Exposed Soil — remained challenging, with F1-scores ranging from 0.31 to 0.52. The comparative analysis of loss functions indicated limited influence on overall performance, reinforcing the robustness of the U-Net architecture. The results highlight the potential of combining CNNs with freely available high-resolution imagery for environmental monitoring in tropical forest regions.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101707"},"PeriodicalIF":4.5,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145159868","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Estimating soil organic carbon using time series Band 11 (SWIR) of multispectral Sentinel-2 satellite images and machine learning algorithms","authors":"Mehdi Golkar Amoli ,&nbsp;Mahdi Hasanlou ,&nbsp;Farhad Samadzadegan ,&nbsp;Ruhollah Taghizadeh-Mehrjardi ,&nbsp;Farzaneh Dadrass Javan","doi":"10.1016/j.rsase.2025.101736","DOIUrl":"10.1016/j.rsase.2025.101736","url":null,"abstract":"<div><div>Soil Organic Carbon (SOC) is a critical soil property impacting food security and climate change. Traditional methods for SOC estimation are time-consuming, expensive, and unsuitable for large-scale application. Consequently, researchers have increasingly focused on utilizing Remote Sensing (RS) images for SOC estimation over the past two decades. However, achieving high SOC estimation accuracy (more than 80 %) remains challenging. This limitation often stems from a mismatch between the complexity of SOC and the information captured by traditional RS observations (e.g., reflectance bands or spectral indices), as conventional feature extraction methods from RS images may not be detailed enough to monitor the many factors influencing SOC concentration. One promising solution to enhance feature extraction is the use of time series observations, analyzing multiple images over time instead of relying on single-time images. This study proposes a novel approach leveraging time series of the Sentinel-2 satellite's B11 band (centered around 1610 nm, a region sensitive to SOC absorption features) along with Principal Component Analysis (PCA) and Independent Component Analysis (ICA) transformations to extract more meaningful temporal features. Specifically, ten new features based on temporal variations were derived by applying PCA and ICA to the B11 band time series images. These temporal features were then combined with features derived from the median of all Sentinel-2 images acquired during the summer of 2019, corresponding to the soil data collection period. Four machine learning algorithms (RF, GBRT, XGBoost, and LightGBM) were employed across four distinct scenarios to evaluate the novel feature extraction method and a feature selection algorithm. The scenarios were designed as follows: Scenario one (S#1) and Scenario two (S#2) did not utilize the time series features, while Scenario three (S#3) and Scenario four (S#4) did. A binary Genetic Algorithm (GA) for feature selection was implemented in S#2 and S#4, distinguishing them from S#1 and S#3 respectively. XGBoost performed best, achieving an R² of 0.891 in S#4 (time series features and GA). Incorporating time series features significantly improved accuracy by 0.11, while GA-based feature selection added another 0.05. The findings highlight the effectiveness of the developed feature extraction algorithm, using Sentinel-2's B11 time series and advanced transformations, for substantially improving SOC level estimation.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"40 ","pages":"Article 101736"},"PeriodicalIF":4.5,"publicationDate":"2025-09-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145159869","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}