Remote Sensing Applications-Society and Environment最新文献

{"title":"Integrating multi-temporal remote sensing and advanced drought modeling to assess desertification dynamics in semi-arid Andhra Pradesh, India: A framework for sustainable Land management","authors":"Pradeep Kumar Badapalli","doi":"10.1016/j.rsase.2025.101654","DOIUrl":"10.1016/j.rsase.2025.101654","url":null,"abstract":"<div><div>This study aims to develop a robust framework for assessing desertification dynamics in the semi-arid landscapes of Andhra Pradesh, India, by integrating multi-temporal remote sensing data with advanced drought modeling. The primary objective is to evaluate the spatiotemporal progression of land degradation by analyzing vegetation response to drought stress over a 30-year period (1990–2020). The Standardized Precipitation Index (SPI) was calculated using RStudio at 3-, 6-, 9-, and 12 - months intervals based on rainfall data derived from CHIRPS satellite-based precipitation, to characterize drought intensity and frequency. Concurrently, Landsat imagery (TM, ETM+, and OLI/TIRS) was processed to generate Normalized Difference Vegetation Index (NDVI) time series to assess vegetation cover changes. A Desertification Status Map (DSM) was prepared by integrating SPI metrics with NDVI-based land cover classifications for the years 1990, 2000, 2010, and 2020. The DSM classified the landscape into four severity categories: Highly Safe (79.45 km²), Safe (248.54 km²), Degraded (320.39 km²), and Desertified Land (402.57 km²). Results highlight a significant increase in degraded and desertified areas, particularly in the western region and along the Hagari River, driven by prolonged drought, vegetation loss, and aeolian activity. Validation of the DSM using 120 ground truth points and high-resolution overlays achieved an overall accuracy of 87.5 % confirming classification reliability. The proposed framework offers a scalable tool for monitoring desertification and supports data-driven planning for sustainable land management, particularly in vulnerable semi-arid ecosystems affected by climate variability and anthropogenic pressures.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101654"},"PeriodicalIF":3.8,"publicationDate":"2025-07-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144589077","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":"Smartphone-based image analysis and interpretable machine learning for soil moisture estimation across diverse Indian soils","authors":"Debasish Roy ,&nbsp;Tridiv Ghosh ,&nbsp;Bappa Das ,&nbsp;Raghuveer Jatav ,&nbsp;Debashis Chakraborty","doi":"10.1016/j.rsase.2025.101655","DOIUrl":"10.1016/j.rsase.2025.101655","url":null,"abstract":"<div><div>Reliable soil moisture estimation is crucial for agricultural water management, yet conventional methods are often invasive, costly, and impractical for frequent field-level use. This study presence a smartphone-based, non-destructive approach for estimating soil moisture content (SMC) estimation across five contrasting Indian soil groups from 14 locations. A total of 238 soil images were analyzed to extract 33 colour-based features, which were then used to train and validate ten machine learning (ML) models. The Random Forest (RF) model exhibited the highest predictive accuracy (R² = 0.78; RMSE = 5.98 %) during validation. To improve interpretability, SHAP and ALE techniques identified Redness Index (RI), Colour Feature Index (ColFeatInd), red band (R), value (V), and X colour space as key predictors. Boruta selection confirmed the relevance of all features. This study demonstrates the potential of combining smartphone imagery and interpretable ML to scalable, low-cost SMC across diverse soil types.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101655"},"PeriodicalIF":3.8,"publicationDate":"2025-07-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144589076","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":"Evaluating satellite data and deep learning for identifying direct deforestation drivers in Cameroon","authors":"Amandine Debus ,&nbsp;Emilie Beauchamp ,&nbsp;Justin Kamga ,&nbsp;Astrid Verhegghen ,&nbsp;Christiane Zébazé ,&nbsp;Emily R. Lines","doi":"10.1016/j.rsase.2025.101653","DOIUrl":"10.1016/j.rsase.2025.101653","url":null,"abstract":"<div><div>Deforestation rates have been increasing in the Congo Basin in recent years, especially in Cameroon. To support actions to slow deforestation, Earth Observation (EO) has been used extensively to detect forest loss, but approaches to automatically identify specific drivers of deforestation in a level of detail that allows for intervention prioritisation (e.g. focusing on specific areas and actions, designing measures to address specific drivers) have been rare. In this paper, using a new country-specific dataset created for this task, we test whether deep learning with optical satellite data can reliably identify direct drivers of deforestation in Cameroon. We compare the effectiveness of two types of freely available optical satellite imagery of differing spatial resolutions: Landsat-8 (30 m) and NICFI PlanetScope (4.77 m). Since it can be challenging to know which collections are best suited for specific applications, we tested different ones to find the optimal approach. Our detailed classification strategy includes fifteen direct deforestation drivers for forest loss events taking place between 2015 and 2020. We obtain a macro-average F1 score of 0.77 with Landsat-8 data, and a macro-average F1 score of 0.65 with NICFI PlanetScope. Despite a coarser spatial resolution, Landsat-8 performs better than NICFI PlanetScope overall, including for small-scale drivers, although results vary by class. Using only a single-image approach, we achieve F1 scores above 0.65 for all classes except ‘Oil palm plantation’, ‘Hunting’ and ‘Fruit plantation’ with Landsat-8. Our results demonstrate the potential of this approach to monitor and analyse land-use changes leading to deforestation with more refined classes than before. Further, our study demonstrates the potential of leveraging existing available datasets and straightforwardly adapting a generalised framework for other regions experiencing rapid deforestation with only a relatively small amount of location-specific data.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101653"},"PeriodicalIF":3.8,"publicationDate":"2025-07-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144605603","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":"Satellite-based mapping and modeling of mangrove loss: A systematic review of significant parameters and methods with an integrated meta-analysis of local ecological knowledge","authors":"Alvin B. Baloloy","doi":"10.1016/j.rsase.2025.101651","DOIUrl":"10.1016/j.rsase.2025.101651","url":null,"abstract":"<div><div>To accurately map and estimate mangrove loss, it is crucial to develop insights from scientific literature, data analytics, and Local Ecological Knowledge (LEK) to characterize the dynamic interplay of anthropogenic impacts, climate change, and conservation efforts. This study presents a meta-analysis of 159 included articles, selected from 1316 publications retrieved through Scopus, Web of Science, and other open-access and journal databases. Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA), the study identifies significant parameters and methodologies applied to satellite data for quantifying mangrove loss and degradation at local, regional, and global scales.</div><div>This parameter-centric meta-analysis identified 76 significant mapping and modeling parameters across categories including optical and Synthetic Aperture Radar (SAR) bands, vegetation and moisture indices, soil and built-up indices, elevation, environmental and temporal dynamics, climatic and geographical factors, proximity metrics, and socio-economic factors. Vegetation indices emerged as the most frequently utilized parameter with 110 articles, highlighting the substantial focus on vegetation metrics in existing research. The highest loss proportions were associated with the use of built-up indices, vegetation indices, SAR bands, and temporal dynamics, ranging between 31 % and 37 %. However, the high standard deviation in vegetation and temporal dynamics suggests the importance of selecting specific indices to accurately capture mangrove degradation. The meta-analysis highlighted a significant research gap in understanding mangrove loss, with only 7.5 % of studies focusing on predictive modeling. Additionally, the supplementary analysis of LEK provided community-perceived factors affecting mangrove loss, offering context-specific insights for effective management. This review highlights the need for more studies on forecasting mangrove loss under various future scenarios, the use of open-source software and adaptive modeling tools, and the integration of LEK and community engagement to enhance the relevance, accuracy, and impact of mangrove conservation strategies.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101651"},"PeriodicalIF":3.8,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144549371","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":"Improving wildfire simulation accuracy using satellite active fire data for interval reinitialization and rate of spread adjustment","authors":"Shahab M. Beyki ,&nbsp;António Manuel Gameiro Lopes ,&nbsp;Aldina Santiago ,&nbsp;Luís Laím","doi":"10.1016/j.rsase.2025.101648","DOIUrl":"10.1016/j.rsase.2025.101648","url":null,"abstract":"<div><div>Wildfires are reoccurring events that burn millions of hectares over the world every year resulting in ecosystem and economic damage and loss of life, and they are becoming more severe and frequent due to climate changes and global warming. Wildfire simulators are fire behavior prediction tools that can be used to manage fires. However, many factors affect the accuracy of simulations and the results are prone to uncertainties. Rate of spread (ROS) adjustment is a method that improves the accuracy of fire spread models by using data on the location and arrival time of actual fires. However, this task used to be time-intensive, prone to errors, and data in remote fire areas were scarce or inconsistent. The required fire arrival time control data points are obtained through ground or aerial operations. Earth observations (EO) data offer valuable, reliable, easily accessible, and freely available means that can be used to bridge this gap. Satellite active fire data is an EO product that presents the spread of fire near real-time and is an effective way to assess and analyze the accuracy of simulations and improve them. This work develops the innovative method of combining data-driven simulation reinitialization using Visible Infrared Imager Radiometer Suite (VIIRS) active fire data with the Wildfire Analyst's (WFA) automated ROS adjustment algorithm to improve the accuracy of simulations. To avoid accumulated errors in wildfire modeling, which increases drastically when fires last long, this method simulates the fire for 12-h intervals aligned with the VIIRS data production, then adjusts the ROS based on the provided satellite data. Five case studies in Portugal were chosen to include a variety of burn durations and fuel type models to assess this method. This approach significantly improved in reducing error and matching the simulated fire ROS to the actual fire, which also led to more accurate simulations for subsequent burning periods. The mean absolute percentage error (MAPE) in the unadjusted simulations was improved from an average of 71.43 % in 5 case studies to 13.99 %. The mean biased percentage error (MBPE) was decreased from 59.12 % on average for case studies to 7.38 %. The accuracy of satellite data and resolution, overpass interval time, affects of environmental factors on the adjustment, and fuel up ahead of the fire that remain unadjusted are the main limitations of this method. This method can be used as a practical approach in real-life incidents for battling and managing fires to increase the accuracy of operation, resource allocation, and decision-making in real time.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101648"},"PeriodicalIF":3.8,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144549369","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 radar and multi-spectral data to detect cocoa crops: a deep learning approach","authors":"Adele Therias ,&nbsp;Azarakhsh Rafiee ,&nbsp;Stef Lhermitte ,&nbsp;Philip van der Lugt ,&nbsp;Roderik Lindenbergh","doi":"10.1016/j.rsase.2025.101652","DOIUrl":"10.1016/j.rsase.2025.101652","url":null,"abstract":"<div><div>The production of cocoa beans contributes to 7.5 % of European Union (EU) driven deforestation. As a result, the recent European Union Deforestation-free Regulation (EUDR) mandates producers to track cocoa farm extents comprehensively. While Remote Sensing has enormous capacity in dynamic crop monitoring, cocoa crop detection shows challenges due to cocoa complex canopy structure, spectral similarity to forest, variable farming methods, and location in frequently cloudy regions. Previous research on cocoa crop detection has mainly focused on pixel-based classification, disregarding spatial context. In this research we have performed a semantic segmentation approach to incorporate spatial configuration and enhance cocoa crop detection. We have applied Convolutional Neural Network (CNN) for the to semantic segmentation of cocoa parcels, considering both spectral and spatial characteristics. Additionally, we have evaluated the impact of combining Synthetic Aperture RADAR (SAR) and MSI (Multi-Spectral Imagery) data in the training of a CNN to demonstrate the importance of texture, moisture, and canopy characteristics in identifying cocoa canopies. The impact of MSI dataset stack with different SAR polarizations, seasons and temporality has been evaluated. The methodology is tested on Sentinel 1 and 2 data over an area of 100 × 100 km in Ghana for which an extensive ground truth data set of almost 90,000 polygons was available for training and validation. The results show that the addition of single-day and temporal SAR to a single-day MSI image can improve the predictions, reaching an F1 score of 86.62 %. This research demonstrates the influence of SAR measurements, seasons, polarization, and ground truth classes on the semantic segmentation of cocoa.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101652"},"PeriodicalIF":3.8,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144534763","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":"Assessment of land deformation and groundwater depletion using Sentinel-1 PS InSAR, GRACE, and Borehole data","authors":"Hamid Gulzar ,&nbsp;Sajid Ghuffar ,&nbsp;Syed Amer Mahmood ,&nbsp;Saif Ullah Akhter ,&nbsp;Hania Arif ,&nbsp;Dmitry E. Kucher ,&nbsp;Aqil Tariq","doi":"10.1016/j.rsase.2025.101639","DOIUrl":"10.1016/j.rsase.2025.101639","url":null,"abstract":"<div><div>This study integrates remote sensing, groundwater monitoring, and geospatial techniques to investigate groundwater dynamics and land deformation in Lahore, Pakistan. PS InSAR technique was utilized from 2014 to 2023 to identify significant land subsidence trends, with 246,843 permanent scatter points produced. Land Deformation, ranging from 0.7 to −4.3 cm/year, was observed, particularly from −2.5 to −4.3 cm/year in densely built-up areas. The analysis revealed at the county-scale (administrative unit) varying levels of subsidence, with Model Town county experiencing the most pronounced deformation with significant subsidence in the Township and its neighboring Union Councils (UCs). Groundwater Storage Anomaly (GWSA) assessments using GRACE Satellite data showed a rapid decline in groundwater storage, particularly in Cell B (−0.77 tau) of the JPL Mascon dataset and across all cells in the CSR Mascon dataset. Depth to Water (DTW) measurements from groundwater monitoring wells indicated significant spatial variations in groundwater levels, with central Lahore exhibiting considerably higher median DTW values than suburban areas. Additionally, correlation analysis using Spearman's method revealed strong associations, particularly in Model Town county, among all GRACE cells (&gt;0.75 ρ), and with DTW in high-subsidence areas. Trend and slope analysis have also been conducted to support these findings further. The study recommends validating InSAR data with Global Navigation Satellite System (GNSS) measurements and GWSA with water table data of small intervals to enhance the accuracy and reliability of land deformation and groundwater assessments. These findings aid policymakers and urban planners in sustainable groundwater management and land use planning in Lahore.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101639"},"PeriodicalIF":3.8,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144570715","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":"Integration of recurrent neural network models and flood inundation extent libraries for flood forecasting","authors":"Vinod Kr. Sharma ,&nbsp;Abhinav Kr. Shukla ,&nbsp;V.M. Chowdary ,&nbsp;Sameer Saran ,&nbsp;S. Kr. Srivastav","doi":"10.1016/j.rsase.2025.101649","DOIUrl":"10.1016/j.rsase.2025.101649","url":null,"abstract":"<div><div>High rainfall events have increased the frequency of floods worldwide, resulting in significant loss of life and property. Developing countries like India face severe flood situations across various states during the monsoon season. Timely and accurate flood forecasting can help disaster management authorities save lives through timely evacuation. The utilisation of deep learning models can aid in accurate flood water level prediction. Recurrent Neural Network like Gated Recurrent Unit (GRU) and Long Short-Term Memory (LSTM) models have potential to analyze sequential data. There is a need to compare available deep learning models, including GRU and LSTM models, to identify the most efficient model for river water level forecasting. This study applies GRU and LSTM models to satellite-derived rainfall, soil moisture, and temperature data, as well as ground-based river water level measurements in the upper river basin, to forecast river water levels in the lower region of the Bagmati river basin. The LSTM models, particularly with the Swish activation function, outperform GRU models in terms of accuracy (89 %), Mean Square Error (MSE, 0.0163), Mean Absolute Error (MAE, 0.0864), and R-squared (0.9630), demonstrating superior predictive capabilities. This work is further enhanced by integrating historical flood extent libraries, derived from remote sensing satellite data and water levels at gauge stations, to simulate probable flood inundation at specific water levels. Comparative analysis of different deep learning models and the integration of flood extent libraries significantly improves the reliability and accuracy of flood forecasting.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101649"},"PeriodicalIF":3.8,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144524120","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":"Combining spectral water index with band for surface water area extraction by using Google Earth Engine (GEE) and ArcGIS in the southern low mountain and hilly areas of China","authors":"Kindeneh Bekele Emiru ,&nbsp;Yin Ren ,&nbsp;Shudi Zuo ,&nbsp;Abiot Molla ,&nbsp;Ayalkibet Mekonnen Seka ,&nbsp;Jiaheng Ju","doi":"10.1016/j.rsase.2025.101650","DOIUrl":"10.1016/j.rsase.2025.101650","url":null,"abstract":"<div><div>Monitoring and evaluating surface water dynamics is crucial for addressing climate change and fostering growth in various sectors. The spectral water index method is a predominant approach for mapping and monitoring surface water. The study was conducted in the southern low mountain and hilly areas of China.</div><div>This study presents a combined approach to enhancing extraction accuracy in surface water mapping. Five distinct water and vegetation indices were employed alongside various bands. The modified normalized difference water index (mNDWI), combined with the near-infrared (NIR) band, has demonstrated superior extraction accuracy across different types of water compared to the other combinations. The combination of (mNDWI_NIR) was validated with the Joint Research Center (JRC) product Global Surface Water (GSW) dataset and the available surface runoff data. The accuracy of the combined method was assessed using a confusion matrix, which yielded an overall accuracy of 96.90 % and a kappa value of 0.868. It also shows a strong linear correlation with areal surface runoff distributions, with an R² value of 0.946, compared to GSW and land use and land cover (LULC) values of 0.933 and 0.926, respectively.</div><div>The method demonstrated a comprehensive approach, stability, and versatility across various environmental conditions over the years in efficiently extracting slender waters. Its usefulness is shown by analyzing spatiotemporal dynamics in the Southern low mountain and hilly areas of China, highlighting its capacity to expand to larger regions, which supports the efforts of the government and water management authorities to recover and restore water resources.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101650"},"PeriodicalIF":3.8,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144685768","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":"Multi-model convolutional neural network architectures for coastal forest extent and aboveground biomass estimation","authors":"Angelo R. Agduma ,&nbsp;Richard Dein D. Altarez","doi":"10.1016/j.rsase.2025.101647","DOIUrl":"10.1016/j.rsase.2025.101647","url":null,"abstract":"<div><div>Mapping coastal forests through large-scale remote sensing remains challenging, despite extensive local, national, and global efforts. In particular, Sarangani Bay Protected Seascape (SBPS) in the Philippines has been largely overlooked in both national and global coastal forest mapping initiatives. To address this gap, we evaluated the performance of three convolutional neural network (CNN) models, U-Net, DeepLabV3, and PSPNet, in identifying coastal forests within SBPS. These forested areas detected were subsequently analyzed for leaf area index (LAI), which was then used to estimate aboveground biomass (AGB). Among the models tested, U-Net demonstrated the highest accuracy, achieving an overall accuracy of 92.66 %. In contrast, DeepLabV3, while the fastest to train, yielded lower accuracy. AGB estimates revealed that the municipalities of <em>Glan</em> and <em>Maasim</em> had the highest AGB, with 2582.43 Mg ha⁻¹ and 1260.57 Mg ha⁻¹, respectively, while <em>Alabel</em> recorded the lowest at 27.27 Mg ha⁻¹. Although distinguishing true mangroves from non-true mangrove classes in coastal forests remains a limitation, the integration of remote sensing and deep learning offers strong potential for enhancing the accuracy and efficiency for land use and land cover classification, as well as AGB estimation.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"39 ","pages":"Article 101647"},"PeriodicalIF":3.8,"publicationDate":"2025-06-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144524291","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}