Remote Sensing Applications-Society and Environment最新文献

{"title":"A metaheuristic design for hyperparameter tuning of XGBoost: Towards shaping an impeccable strategy for predictive modeling of mineral prospectivity","authors":"Saeid Esmaeiloghli ,&nbsp;Mahyar Yousefi ,&nbsp;Emmanuel John M. Carranza","doi":"10.1016/j.rsase.2026.101980","DOIUrl":"10.1016/j.rsase.2026.101980","url":null,"abstract":"<div><div>Over the past two decades, machine learning (ML) has been an avant-garde technology for data-driven mineral prospectivity mapping (MPM) due to its competence in modeling non-linear relationships embedded in multi-source geoscience datasets. XGBoost is a flagship algorithm in boosting-based ML that has gained popularity in MPM for its proficiency in speeding up the training process, enhancing prediction accuracy, reducing the risk of overfitting, and improving generalizability. However, achieving a well-trained XGBoost model usually entails careful tuning of multiple hyperparameters, i.e., <em>n_estimators</em>, <em>max_depth</em>, <em>min_child_weight</em>, <em>gamma</em>, <em>subsample</em>, <em>colsample_bytree</em>, and <em>learning_rate</em>. Motivated by this challenge, we conceptualized a computational framework in this paper that employs particle swarm optimization (PSO) to discover optimal XGBoost-related hyperparameters that yield MPM predictions with the highest accuracy. The PSO algorithm was engineered using a training dataset and a five-fold cross-validation strategy to find a hyperparameter setting that is globally optimal for achieving an XGBoost model with a robust performance. The PSO‒XGBoost model was coded and implemented by scripting over functions developed within the R programming language. The potential application of the proposed hybrid model was demonstrated through a real-case experiment for predictive modeling of porphyry-type copper prospectivity in the Baft-Sarduiyeh district, southern Iran. A comparative analysis between the PSO‒XGBoost model and manually tuned XGBoost scenarios revealed the superiority of the former by delivering higher values for confusion matrix-derived evaluation metrics and creating a higher-performance curve during receiver operating characteristic analysis. The findings suggest that tuning XGBoost with PSO-optimized hyperparameters can significantly improve predictive power and promote interpretability for more accurate modeling of mineral prospectivity.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 101980"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147613480","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":"Relationships between spectral and biological diversity depend on season and habitat type","authors":"Soroor Rahmanian ,&nbsp;Nico Eisenhauer ,&nbsp;Antonia Ludwig ,&nbsp;Yuanyuan Huang ,&nbsp;Hannes Feilhauer","doi":"10.1016/j.rsase.2026.102039","DOIUrl":"10.1016/j.rsase.2026.102039","url":null,"abstract":"<div><div>Remote sensing is increasingly used to monitor biodiversity, with spectral diversity—pixel-to-pixel variation in spectral reflectance—serving as a key proxy for taxonomic and functional diversity. However, seasonal dynamics and ecological drivers underlying spectral–biological diversity relationships remain less understood. This study examines seasonal patterns across three temperate open habitats in Germany—a nutrient-poor grassland, wet heathland, and floodplain meadow. We monitored 130 1 m² plots over three seasons, measuring taxonomic diversity (Shannon, Simpson, inverse Simpson, Pielou's evenness, and species richness), functional diversity (functional dispersion, richness, evenness, divergence, Rao's Q), and four spectral diversity indices (average angle dissimilarity, coefficient of variation of whole spectra and optical traits, RaoQ) across narrow 166 wavelength regions, along with vegetation parameters. Data were collected on six to seven dates during the growing season using a field spectrometer to capture seasonal and trait variation. We employed linear mixed-effects and structural equation models to evaluate how spectral diversity reflects biodiversity over time and across habitats. Results suggested that these relationships vary across habitats and seasons. Vegetation structure — especially non-photosynthetic vegetation (NPV; senescent/litter) and canopy height — are linked to spectral and biological diversity through context-dependent pathways that vary across habitats and seasons. NPV was positively associated with spectral diversity in the grassland and floodplain, whereas canopy height showed temporally variable effects, enhancing functional diversity mid-season but exhibiting weaker or negative relationships in the heathland. These findings highlight that links between spectral and biological diversity are context dependent and vary with vegetation structure and season.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102039"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147802874","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 weighted ensemble DL framework for granular RPV inventory and energy yield analysis in Pakistan","authors":"Muhammad Kamran Lodhi ,&nbsp;Yumin Tan ,&nbsp;Yang Li ,&nbsp;Faisal Mumtaz ,&nbsp;Shahid Naeem","doi":"10.1016/j.rsase.2026.102032","DOIUrl":"10.1016/j.rsase.2026.102032","url":null,"abstract":"<div><div>The global imperative to triple renewable energy by 2030 underscores the vital role of Rooftop Photovoltaics (RPV). Despite high solar irradiation, Pakistan's RPV potential remains underexplored due to the poor generalizability of conventional Deep Learning (DL) models across complex urban architectures. This study addresses this limitation by proposing a weighted ensemble DL framework for large-scale RPV delineation in Islamabad, Lahore, and Karachi. Multiple DL models were trained on region-specific, high-resolution Google Satellite imagery (0.3 m spatial resolution), with final predictions synthesized through a performance-based weighted majority voting scheme. For solar resource assessment, a 10 m Digital Surface Model (DSM) was generated from Sentinel-1 data to account for urban morphology. The ensemble model demonstrated superior robustness, achieving an F1-score of 0.92, 0.96 accuracy, and a Matthews Correlation Coefficient of 0.89. Applying this framework, we identified 21,586 installations in Islamabad, 65,304 in Lahore, and 35,710 in Karachi, representing the most granular assessment to date. Analysis of annual electricity yield and carbon mitigation revealed that Karachi leads with 602 GWh (0.37 Mt CO₂ reduction), followed by Lahore at 373 GWh (0.23 Mt CO₂), and Islamabad at 141 GWh (0.09 Mt CO₂). Total annual generation from identified modules reached 1117.94 GWh, offsetting 0.693 Mt CO₂. Ground validation in Lahore confirmed high model reliability (R² = 0.98, MAPE = 6.02%). This research provides a scalable, data-driven methodology for RPV mapping in diverse developing regions, offering essential intelligence for policymakers to accelerate Pakistan's transition to a sustainable energy future.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102032"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147802876","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 U-Net with enhanced spatiotemporal features for Chlorophyll-A concentration prediction in the Zhoushan fishery","authors":"Yanling Han,&nbsp;Yizhuo Cai,&nbsp;Ping Ma,&nbsp;Ge Song,&nbsp;Jing Wang,&nbsp;Yun Zhang,&nbsp;Shuhu Yang,&nbsp;Zhonghua Hong","doi":"10.1016/j.rsase.2026.101986","DOIUrl":"10.1016/j.rsase.2026.101986","url":null,"abstract":"<div><div>Chlorophyll-a (Chl-a) is a key indicator of marine ecological conditions, and its abrupt increase often signals the occurrence of harmful algal blooms (HABs). As one of China's major fishing grounds, the Zhoushan fishing ground is highly susceptible to HAB events. However, existing Chl-a prediction methods, which commonly rely on single-source or limited fixed-site observations, often suffer from insufficient integration of spatiotemporal information, inadequate exploration of intrinsic feature correlations, and limited capability in capturing the spatiotemporal dependencies and variability of Chl-a distributions. These limitations reduce their robustness, particularly under extreme meteorological disturbances. To address these challenges, this study proposes a Spatiotemporal Enhanced U-Net with Convolutional Long Short-Term Memory (STE-U-ConvLSTM) for daily Chl-a prediction in the Zhoushan fishing ground based on satellite remote sensing data. Specifically, an enhanced spatiotemporal feature representation strategy is developed by incorporating temporal and geographic encodings, Chl-a concentration gradients, and directional angles to better characterize Chl-a dynamics. Built upon the U-Net architecture, the proposed model embeds a Convolutional Block Attention Module (CBAM) into the encoder to adaptively emphasize informative spatiotemporal features. In addition, the bottleneck layer integrates dilated convolutions with ConvLSTM to capture multiscale spatiotemporal patterns more effectively. Experimental results show that STE-U-ConvLSTM outperforms baseline models, increasing R² by up to 0.139 and reducing RMSE by up to 0.25 μg/L relative to baseline models. The proposed model also demonstrates strong generalization across different seasons and subregions, with particularly notable performance in complex summer and winter conditions as well as dynamic nearshore waters, where external forcing plays a dominant role. Furthermore, the model proves effective for the early warning of actual HAB events, providing a reliable tool for marine environmental monitoring, bloom forecasting, and sustainable fishery management.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 101986"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147613004","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 novel index for mapping plastic-mulched farmland using multi-temporal Landsat-8 and Sentinel-2 imagery","authors":"Junyi Chang ,&nbsp;Qian Yang ,&nbsp;Feng Tao ,&nbsp;Huanjun Liu ,&nbsp;Jia Du ,&nbsp;Chong Luo ,&nbsp;Yuan Chai ,&nbsp;Liwen Chen ,&nbsp;Xue Li","doi":"10.1016/j.rsase.2026.101992","DOIUrl":"10.1016/j.rsase.2026.101992","url":null,"abstract":"<div><div>Plastic mulching is a prevalent technology in protected cultivation, and remote sensing offers a robust tool for monitoring the spatiotemporal dynamics of plastic-mulched farmland (PMF). However, extracting PMF information using remote sensing presents several challenges. These challenges include the similarity in spectral characteristics between agricultural films and other land covers, as well as the difficulty in achieving accurate classification over large areas with varying conditions. Traditional indices (such as NDVI, NDWI, NDBI, EVI) are not suitable for detecting PMF because they lack sensitivity to the unique spectral signatures of plastic mulch. Therefore, we proposed the Normalized Agricultural Film Index (NAFI), which has been applied to Landsat 8 OLI and Sentinel 2 MSI images acquired in April and May through Google Earth Engine (GEE). This index has demonstrated superior performance in PMF identification. We designed five different feature combinations, selected the optimal combination based on classification accuracy and performance metrics, and performed the Random Forest (RF) to map PMF. The overall classification accuracy of RF ranges from 92.04% to 99.27%, with kappa coefficients between 0.87 and 0.98, highlighting the effectiveness of the proposed approach in large-scale agricultural monitoring. This study monitors the spatial and temporal changes of agricultural film from 2015 to 2023 in Da’ an, Jilin Province, Northeast China, which witness a northwest expansion.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 101992"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147613378","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":"Smart integration of sliding window and vote-based fusion: Advancing UAV-based instance segmentation with YOLOv8 for high-resolution vegetation mapping","authors":"Ilias Jennaoui ,&nbsp;El Mostafa Bachaoui ,&nbsp;Mohamed Biniz ,&nbsp;Abderrazak El Harti ,&nbsp;Abdrrahmane El Ghmari","doi":"10.1016/j.rsase.2026.101994","DOIUrl":"10.1016/j.rsase.2026.101994","url":null,"abstract":"<div><div>Vegetation segmentation from high-resolution drone imagery faces fundamental challenges due to scale mismatches between training and inference phases. This issue becomes particularly pronounced when limited high-resolution images are available for model training, necessitating patch-based approaches that create significant scale inconsistencies. The present study addresses these challenges by introducing a spatial consensus voting methodology with confidence-weighted fusion for high-resolution vegetation mapping in the El Ksiba region of Morocco. Unlike previous multi-model ensemble approaches, the proposed method employs a single model generating 10-12 overlapping predictions per pixel, achieving 8.2 × performance recovery from scale mismatch while requiring 3 × less computational cost. Comprehensive evaluation across multiple architectures demonstrates that DeepLabV3+ provides superior precision (0.882) and boundary coherence, while YOLOv8 delivers high recall (0.923) suitable for comprehensive coverage. Contrary to conventional expectations, class imbalance methods consistently degrade deployment performance despite training improvements. This research reveals a focal loss paradox wherein +14.1% training improvement reverses to −1.7% deployment degradation through complex interactions with deployment parameters. Furthermore, minority species benefit substantially more from scale-consistent inference (+55.9% for <em>Lentisque</em>) than from training-time balancing approaches. Experimental results demonstrate effectiveness with mean Intersection over Union of 0.721 and F1-score of 0.826, providing both comprehensive vegetation detection and reliable classification. By resolving scale mismatch problems inherent in training on resized images, this study presents a scalable solution for vegetation segmentation in remote sensing applications, with potential extensions to multimodal sensor integration for enhanced species-specific classification.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 101994"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147613481","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":"The investigation of variations in chlorophyll-a in the southern coast of Sulawesi: a case study from 1998 to 2023","authors":"Yusuf Jati Wijaya ,&nbsp;Ulung Jantama Wisha ,&nbsp;Bayu Munandar ,&nbsp;Lilik Maslukah ,&nbsp;Seto Windarto ,&nbsp;Muhammad Zainuri","doi":"10.1016/j.rsase.2026.102016","DOIUrl":"10.1016/j.rsase.2026.102016","url":null,"abstract":"<div><div>The southern coast of Sulawesi has the potential to produce yearly upwelling events, exhibiting significant seasonal variations in chlorophyll-a (Chl-a) concentration. However, the mechanism responsible for the seasonal fluctuations of Chl-a has not been entirely revealed. Utilization of the latest long-term data to ascertain annual variability patterns also remains infrequent. Therefore, this study employed a combination of Chl-a satellite data and reanalysis data of oceanographic parameter to explore the factors influencing seasonal and annual fluctuations in upwelling event. We found that the seasonal rise in Chl-a was most pronounced during the peak intensity of the southeasterly monsoon (SEM), specifically in June, July, and August (JJA). In the year 2004, Chl-a concentrations reached their peak during the period of 1998-2023, especially in August, averaging 1.6 mg/m³. This occurred concurrently with the coldest water temperature in that year, which fell at 26.3 °C, in comparison to other years. Despite the fact that 2004 exhibited weak El Niño events and a neutral Indian Ocean Dipole (IOD). In 2015, during an extreme El Niño event, and in 2019, characterized by a strong positive IOD, the aquatic conditions exhibited warmer temperatures (26.8° and 26.5 °C) and lower Chl-a concentrations (1.4 and 1.0 mg/m³). We discovered that the significant increase in Chl-a concentration and coldest temperature in 2004 was affected by subsurface ocean current off the western coast and westward flowing surface currents along the southern coast of Sulawesi. The interplay of these two ocean currents significantly influences the variability and spatial distribution of Chl-a and sea surface temperature.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102016"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147710208","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":"Comparative assessment of vegetation cover and management factor in soil loss modelling using Sentinel-2 MSI time-series imagery: A case study from the Romanian Carpathian-Subcarpathian region","authors":"Marina Vîrghileanu ,&nbsp;Ionuț Șandric ,&nbsp;Bogdan-Andrei Mihai ,&nbsp;Ionuț Săvulescu ,&nbsp;Carmen-Gabriela Bizdadea","doi":"10.1016/j.rsase.2026.102013","DOIUrl":"10.1016/j.rsase.2026.102013","url":null,"abstract":"<div><div>Vegetation cover and management (C-factor) is a key component of the RUSLE model, which is widely used for predicting soil erosion. The C-factor is the most dynamic variable, being influenced by plant phenology which introduces substantial uncertainties in erosion modelling. Achieving accurate C-factor mapping is crucial for reliable soil erosion estimation across different scales. The aim of our paper is to assess the performance of complementary approaches for C-factor estimation and analysing of its spatio-temporal variability, using spectral and biophysical indices derived from time-series Sentinel-2 MSI imagery. The study focuses on a test area within the Carpathian-Subcarpathian region of Romania, with potential for the results to be up-scaled to regional or national levels. The workflow involves four key stages: (1) processing Sentinel-2 MSI time-series from 2016 to 2024 to derive spectral and biophysical indices, (2) estimating the C-factor through different remote sensing approaches proposed in literature, (3) analysing the spatio-temporal variability of the C-factor, as well as inter-comparison with empirical values from literature, and (4) estimating seasonal and annual soil loss rates using the RUSLE equation. Our results indicate that the integration of multitemporal imagery from the same sensor enhances the objectivity and the significance of the C-factor estimation at both local and regional levels. The importance of the comparative approach is of practical value in improving the soil loss models from their static to a dynamic framework. The study provides foundational insights for sustainable land-use management scenarios, supporting policy-making and land-use planning.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102013"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147710292","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":"Mapping shrub and tree encroachment in Canadian Prairies using stacking ensemble and Sentinel-1/2 imagery","authors":"Yihan Pu,&nbsp;Irini Soubry,&nbsp;Xulin Guo","doi":"10.1016/j.rsase.2026.102041","DOIUrl":"10.1016/j.rsase.2026.102041","url":null,"abstract":"<div><div>Woody plant encroachment (WPE) threatens grassland ecosystems across the Canadian Prairies, causing grassland biodiversity loss with substantial economic impacts due to reduced forage production. While remote sensing offers scalable monitoring capabilities, existing approaches lack frameworks for distinguishing shrub and tree encroachment and often require extensive ground truth data. This study developed an ensemble machine learning framework integrating Sentinel-1 SAR and Sentinel-2 optical imagery with UAV-derived training data to map fractional shrub and tree cover across Saskatchewan's Aspen Parkland and Moist Mixed Grassland ecoregions, SK. A stacking ensemble combining Random Forest, Support Vector Machine, XGBoost, and Artificial Neural Network models with Ridge regression meta-learning outperformed individual algorithms, achieving mean R² values of 0.65 for shrub and 0.68 for tree cover prediction. Multi-scale training incorporating features at 10, 30, 50, and 70 m resolution improved performance by 15% for shrub and 24% for tree compared to single-scale approaches. Feature importance analysis revealed that shrub detection relied primarily on red-edge bands and moisture indices, while tree detection depended heavily on SAR backscatters. Quantile histogram matching enabled successful model transfer from Foam Lake Community pasture to Aberdeen Community Pasture, with resulting maps indicating that total WPE exceeded 50% in both study areas, with shrubs occupying 23.7% (Foam Lake) and 18.5% (Aberdeen) of both regions at rates higher than 5% shrub cover. The present framework provides a scalable, cost-effective approach for operational woody encroachment monitoring, enabling early detection and targeted functional management interventions to preserve grassland ecosystems.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102041"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147803463","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 globally applicable framework for earthquake precursor identification using gravity gradient tensor anomalies from satellite gravimetry","authors":"Sarfaraz Newaz ,&nbsp;Jannatun Noor ,&nbsp;A.B.M. Alim Al Islam","doi":"10.1016/j.rsase.2026.102000","DOIUrl":"10.1016/j.rsase.2026.102000","url":null,"abstract":"<div><div>Earthquakes remain one of the most unpredictable natural disasters, yet forecasting such events is critical for mitigating their devastating impact. In this study, we introduce a novel framework for earthquake forecasting by analyzing satellite remote sensing data. Specifically, we use Gravity Gradient Tensor (GGT) anomalies derived from the GRACE satellite mission as a prospective pre-seismic precursor. Unlike traditional forecasting methods that are often limited to point-specific epicentral data or complex machine learning ‘black-box’ models, our proposed framework offers the distinct advantage of physical interpretability and continuous global scalability. Our work directly addresses a key debate in the literature by demonstrating the statistical significance of these anomalies, thereby moving beyond previous isolated case studies. Our proposed method is physically interpretable and operates on a continuous, global scale.</div><div>By analyzing historical GRACE data from 2002 to 2016, we developed a unique threshold-based methodology to extract GGT anomalies across five distinct tectonic regions: Japan, Sumatra, Chile, China, and the India–Nepal belt. The framework employs robust statistical filtering to minimize noise and relies on a regionalized approach to establish a dynamic, data-driven threshold. Our results demonstrate high predictive accuracy, with a regional performance of up to 96% true positives in a highly seismic region and as low as 0% false positives. Our method achieves an average true prediction rate of 91% and an average false alarm rate of 13% across all regions. Quantitative comparisons confirm that our model provides superior reliability and significantly lower false alarm rates compared to conventional machine learning classifiers. This research confirms the potential of GGT anomalies as a valuable precursor and establishes a scalable, data-driven framework for global earthquake forecasting, highlighting a new frontier for satellite geodesy applications in solid Earth science.</div></div>","PeriodicalId":53227,"journal":{"name":"Remote Sensing Applications-Society and Environment","volume":"42 ","pages":"Article 102000"},"PeriodicalIF":4.5,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147657707","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}