Accurate seasonal rice yield estimation across Southeast Asia’s intensive cropping systems remains challenging due to complex phenological patterns and heterogeneous environmental conditions. This study develops a phenology-aligned multi-task temporal fusion (MTTF) framework for satellite-based seasonal rice yield estimation in Vietnam’s triple-cropping systems from 2001 to 2020. The multi-task learning treats each cropping season (winter–spring, summer–autumn, monsoon) as related but distinct tasks, enabling knowledge sharing while preserving season-specific characteristics. The framework integrates multi-source time-series data, including climate variables (e.g., temperature, precipitation), satellite-based vegetation indices (e.g., NDVI, EVI, NIRv, GCVI, LSWI), productivity indicators (e.g., SIF, GPP), and static soil properties (e.g., clay content, organic carbon, bulk density) through parallel Transformer encoders and late fusion strategies. To address temporal misalignment across heterogeneous cropping calendars, we developed an automated phenology-based crop season detection method that synchronizes time-series inputs to key growth stages rather than calendar dates. MTTF achieved high performance (R2 = 0.75, RMSE = 0.63 Mg·ha−1, and rRMSE = 12.0%), outperforming baseline models including Transformer, AtBiLSTM, ANN, XGBoost, and Random Forest. The multi-task learning approach outperformed both global models (single predictor for all seasons) and local models (separate predictors for each season), demonstrating particular benefits for data-scarce seasons like monsoon rice. Phenology alignment enhanced temporal consistency across all models. Multi-modal data fusion significantly improved performance, with satellite-based vegetation measurements contributing more significantly than climate variables according to SHAP analysis. The proposed framework provides a robust approach for operational rice yield monitoring across intensive cropping systems, with implications for assessing food security and agricultural policy in monsoon regions.
{"title":"Phenology-Aligned multi-task temporal fusion framework for satellite-based triple-seasonal rice yield estimation in Southeast Asia","authors":"Zhixian Lin, Kaiyu Guan, Sheng Wang, Qu Zhou, Liangzhi You, Xuan Chen, Xiangzhong Luo, Kejie Zhao","doi":"10.1016/j.jag.2026.105231","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105231","url":null,"abstract":"Accurate seasonal rice yield estimation across Southeast Asia’s intensive cropping systems remains challenging due to complex phenological patterns and heterogeneous environmental conditions. This study develops a phenology-aligned multi-task temporal fusion (MTTF) framework for satellite-based seasonal rice yield estimation in Vietnam’s triple-cropping systems from 2001 to 2020. The multi-task learning treats each cropping season (winter–spring, summer–autumn, monsoon) as related but distinct tasks, enabling knowledge sharing while preserving season-specific characteristics. The framework integrates multi-source time-series data, including climate variables (e.g., temperature, precipitation), satellite-based vegetation indices (e.g., NDVI, EVI, NIRv, GCVI, LSWI), productivity indicators (e.g., SIF, GPP), and static soil properties (e.g., clay content, organic carbon, bulk density) through parallel Transformer encoders and late fusion strategies. To address temporal misalignment across heterogeneous cropping calendars, we developed an automated phenology-based crop season detection method that synchronizes time-series inputs to key growth stages rather than calendar dates. MTTF achieved high performance (R<ce:sup loc=\"post\">2</ce:sup> = 0.75, RMSE = 0.63 Mg·ha<ce:sup loc=\"post\">−1</ce:sup>, and rRMSE = 12.0%), outperforming baseline models including Transformer, AtBiLSTM, ANN, XGBoost, and Random Forest. The multi-task learning approach outperformed both global models (single predictor for all seasons) and local models (separate predictors for each season), demonstrating particular benefits for data-scarce seasons like monsoon rice. Phenology alignment enhanced temporal consistency across all models. Multi-modal data fusion significantly improved performance, with satellite-based vegetation measurements contributing more significantly than climate variables according to SHAP analysis. The proposed framework provides a robust approach for operational rice yield monitoring across intensive cropping systems, with implications for assessing food security and agricultural policy in monsoon regions.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"9 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147465120","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Leaf area index (LAI) is a crucial biophysical parameter to characterize the canopy structure and energy absorption capacity of vegetation in Earth system science. While MODIS LAI products are widely used in ecological research, their retrieval algorithms process each pixel and data independently, resulting in high noise levels and time series discontinuities that limit their application. In this study, a Three-step Spatio-temporal Constrained Gap Filling (TSCGF) method for generating high-quality MODIS LAI products in real time is proposed. Our approach consists of three main steps: first, the Spatial Similarity-based Gap Filling (SSGF) uses local neighborhood information within a 50 km × 50 km window to identify similar pixels based on phenological patterns and land cover types, and implements a correlated weighted average scheme to generate a complete LAI time series. Next, globally representative sampling pixels are generated by performing time-series K-means clustering on the pre-filled LAI data within each MODIS tile. Finally, a bidirectional time Convolutional network (BiTCN) model is trained on globally distributed samples, taking advantage of these rich time series to reconstruct high-quality LAI values by capturing long-term time dependencies while maintaining computational efficiency. Since 2000, this method has implemented real-time MODIS LAI datasets on a global scale. Validation of DIRECT2.1 (100 sites, 231 measurements) and GBOV (52 sites, 8423 measurements) ground measurements for different biomes showed that our approach achieved better performance compared to the original MODIS LAI product, with a decrease in root mean square error from 0.95 to 0.88. In areas with frequent cloud cover and complex terrain, TSCGF significantly enhanced spatio-temporal continuity while maintaining true vegetation dynamics, further demonstrating its effectiveness. This approach provides a robust framework for real-time generation of high-quality LAI products that can better support a variety of ecological applications.
叶面积指数（LAI）是地球系统科学中表征植被冠层结构和能量吸收能力的重要生物物理参数。虽然MODIS LAI产品广泛应用于生态研究，但其检索算法对每个像素和数据进行独立处理，导致高噪声水平和时间序列不连续，限制了其应用。本文提出了一种实时生成高质量MODIS LAI产品的三步时空约束间隙填充（TSCGF）方法。该方法包括三个主要步骤：首先，基于空间相似性的间隙填充（SSGF）利用50 km × 50 km窗口内的局部邻域信息，基于物候模式和土地覆盖类型识别相似像元，并实现相关加权平均方案，生成完整的LAI时间序列；接下来，通过对每个MODIS tile内预填充的LAI数据执行时间序列K-means聚类，生成具有全局代表性的采样像素。最后，在全局分布的样本上训练双向时间卷积网络（BiTCN）模型，利用这些丰富的时间序列，在保持计算效率的同时，通过捕获长期时间依赖关系来重建高质量的LAI值。自2000年以来，该方法在全球范围内实现了实时MODIS LAI数据集。对不同生物群落的DIRECT2.1（100个站点，231次测量）和GBOV（52个站点，8423次测量）地面测量结果的验证表明，与原始MODIS LAI产品相比，我们的方法取得了更好的性能，均方根误差从0.95降低到0.88。在云量频繁、地形复杂的地区，TSCGF在保持植被真实动态的同时显著增强了时空连续性，进一步证明了其有效性。这种方法为实时生成高质量的LAI产品提供了一个健壮的框架，可以更好地支持各种生态应用。
{"title":"Real-time generation of gap-free MODIS leaf area index product from 2000 to 2024 using a deep learning method","authors":"Guodong Zhang, Yimin Ni, Gang Sun, Gaofei Yin, Wei Zhao, Anxin Ding, Xinyan Liu, Yi Zhang, Jiangchuan Hu, Zongyan Li, Rui Chen, Meilian Wang, Aleixandre Verger","doi":"10.1016/j.jag.2026.105240","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105240","url":null,"abstract":"Leaf area index (LAI) is a crucial biophysical parameter to characterize the canopy structure and energy absorption capacity of vegetation in Earth system science. While MODIS LAI products are widely used in ecological research, their retrieval algorithms process each pixel and data independently, resulting in high noise levels and time series discontinuities that limit their application. In this study, a Three-step Spatio-temporal Constrained Gap Filling (TSCGF) method for generating high-quality MODIS LAI products in real time is proposed. Our approach consists of three main steps: first, the Spatial Similarity-based Gap Filling (SSGF) uses local neighborhood information within a 50 km × 50 km window to identify similar pixels based on phenological patterns and land cover types, and implements a correlated weighted average scheme to generate a complete LAI time series. Next, globally representative sampling pixels are generated by performing time-series K-means clustering on the pre-filled LAI data within each MODIS tile. Finally, a bidirectional time Convolutional network (BiTCN) model is trained on globally distributed samples, taking advantage of these rich time series to reconstruct high-quality LAI values by capturing long-term time dependencies while maintaining computational efficiency. Since 2000, this method has implemented real-time MODIS LAI datasets on a global scale. Validation of DIRECT2.1 (100 sites, 231 measurements) and GBOV (52 sites, 8423 measurements) ground measurements for different biomes showed that our approach achieved better performance compared to the original MODIS LAI product, with a decrease in root mean square error from 0.95 to 0.88. In areas with frequent cloud cover and complex terrain, TSCGF significantly enhanced spatio-temporal continuity while maintaining true vegetation dynamics, further demonstrating its effectiveness. This approach provides a robust framework for real-time generation of high-quality LAI products that can better support a variety of ecological applications.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"31 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147465202","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-14DOI: 10.1016/j.jag.2026.105160
Yingying Pei, Qunshan Zhao, Junye Wang, Zhi Zheng, Xinyi Liu, Xuejun Huang, Yongjun Zhang, Yi Wan
Rapid urbanization necessitates accurate 3D building data for effective urban planning and analysis. Building height provides critical vertical information reflecting urban morphology, land use intensity, and energy demand. However, high-resolution large-scale 3D datasets remain limited, particularly in the Global South, due to the high cost and complexity of traditional methods. Although machine learning approaches have been widely explored, they often underperform in dense urban areas and tend to underestimate tall buildings due to limited training data and generalizability. In this study, we propose a novel framework for large-scale building height estimation using only free remote sensing data. Leveraging 0.5 m resolution open-access satellite imagery and ICESat-2 photons, we construct an accurate parallel projection model for each image. It enables the generation of dense height points via triangulation across image pairs without additional geometric parameters. The height points are then integrated with 2D building footprints to reconstruct building height maps. Validation results of the full urban area in Nairobi, achieving a root mean square error (RMSE) of 3.338 m, demonstrated the feasibility of our framework. The method also exhibits strong temporal consistency, with a maximum mean deviation of only 1.93 m across multi-temporal height maps. Experiment results in the three additional Global South cities (Medellín, Salvador and Jakarta), achieving mean absolute errors (MAE) of 3.867 m, 3.642 m, and 2.484 m, respectively, further confirmed the transferability of our framework. These results highlight our framework’s capability to deliver low-cost, accurate, and high-resolution 3D urban reconstruction, particularly in resource-constrained cities, providing a scalable tool for urban analysis, planning, and policy support.
{"title":"Three-dimensional time series building reconstruction framework in global south based on openly available satellite data","authors":"Yingying Pei, Qunshan Zhao, Junye Wang, Zhi Zheng, Xinyi Liu, Xuejun Huang, Yongjun Zhang, Yi Wan","doi":"10.1016/j.jag.2026.105160","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105160","url":null,"abstract":"Rapid urbanization necessitates accurate 3D building data for effective urban planning and analysis. Building height provides critical vertical information reflecting urban morphology, land use intensity, and energy demand. However, high-resolution large-scale 3D datasets remain limited, particularly in the Global South, due to the high cost and complexity of traditional methods. Although machine learning approaches have been widely explored, they often underperform in dense urban areas and tend to underestimate tall buildings due to limited training data and generalizability. In this study, we propose a novel framework for large-scale building height estimation using only free remote sensing data. Leveraging 0.5 m resolution open-access satellite imagery and ICESat-2 photons, we construct an accurate parallel projection model for each image. It enables the generation of dense height points via triangulation across image pairs without additional geometric parameters. The height points are then integrated with 2D building footprints to reconstruct building height maps. Validation results of the full urban area in Nairobi, achieving a root mean square error (RMSE) of 3.338 m, demonstrated the feasibility of our framework. The method also exhibits strong temporal consistency, with a maximum mean deviation of only 1.93 m across multi-temporal height maps. Experiment results in the three additional Global South cities (Medellín, Salvador and Jakarta), achieving mean absolute errors (MAE) of 3.867 m, 3.642 m, and 2.484 m, respectively, further confirmed the transferability of our framework. These results highlight our framework’s capability to deliver low-cost, accurate, and high-resolution 3D urban reconstruction, particularly in resource-constrained cities, providing a scalable tool for urban analysis, planning, and policy support.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"95 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209284","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-14DOI: 10.1016/j.jag.2026.105167
Taichang Liu, Fang Chen, Bo Yu, Lei Wang, Mengjie Gao
Post-fire vegetation recovery in boreal forests is shaped by the complex interactions among burn severity, ecological conditions, and climatic factors. Traditional pixel-based assessments often overlook the spatial configuration and shape of burned areas, potentially underestimating heterogeneity and misrepresenting recovery dynamics. Here, we developed a patch-level framework to quantify recovery time for the 2003 Siberian wildfire, examining the combined effects of patch morphology, burn severity, climatic factors, and topography. By 2023, 72.7% of burned patches had recovered, with an average recovery time of 12.2 years. Recovery was fastest in smaller, more compact patches, whereas larger and elongated patches recovered more slowly. Within patches, recovery time increased from edges toward interiors, indicating consistent edge-to-interior gradients. Climatic factors, particularly higher temperature and radiation, slowed recovery, whereas greater precipitation and soil moisture accelerated it. Burn severity also delayed regeneration, while topographic influences were minor in this low-relief landscape. These results demonstrate that patch morphology and climate jointly shape post-fire recovery and emphasize the importance of incorporating spatial configuration into forest resilience assessments and restoration planning. The patch-based framework provides a scalable, ecologically realistic approach applicable to fire-prone ecosystems worldwide.
{"title":"Patch-based assessment of post-fire recovery after the 2003 Siberian wildfire","authors":"Taichang Liu, Fang Chen, Bo Yu, Lei Wang, Mengjie Gao","doi":"10.1016/j.jag.2026.105167","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105167","url":null,"abstract":"Post-fire vegetation recovery in boreal forests is shaped by the complex interactions among burn severity, ecological conditions, and climatic factors. Traditional pixel-based assessments often overlook the spatial configuration and shape of burned areas, potentially underestimating heterogeneity and misrepresenting recovery dynamics. Here, we developed a patch-level framework to quantify recovery time for the 2003 Siberian wildfire, examining the combined effects of patch morphology, burn severity, climatic factors, and topography. By 2023, 72.7% of burned patches had recovered, with an average recovery time of 12.2 years. Recovery was fastest in smaller, more compact patches, whereas larger and elongated patches recovered more slowly. Within patches, recovery time increased from edges toward interiors, indicating consistent edge-to-interior gradients. Climatic factors, particularly higher temperature and radiation, slowed recovery, whereas greater precipitation and soil moisture accelerated it. Burn severity also delayed regeneration, while topographic influences were minor in this low-relief landscape. These results demonstrate that patch morphology and climate jointly shape post-fire recovery and emphasize the importance of incorporating spatial configuration into forest resilience assessments and restoration planning. The patch-based framework provides a scalable, ecologically realistic approach applicable to fire-prone ecosystems worldwide.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"59 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209253","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-14DOI: 10.1016/j.jag.2026.105165
Ruiyang Xu, Rui Yuan, Jing Chen
Estuarine fronts represent critical submesoscale oceanographic phenomena that govern material transport, pollutant dispersion, and ecosystem dynamics in coastal environments. In this study, we developed a comprehensive remote sensing framework for monitoring suspended sediment fronts and plume fronts in the Yangtze River Estuary–Hangzhou Bay transition zone, a region of exceptional ecological and economic significance. Leveraging the Google Earth Engine cloud computing platform, we developed an integrated multi-algorithm approach that combines spectral index enhancement, automated thresholding, clustering analysis, and edge detection techniques for high-precision front feature extraction from Landsat-8 and Sentinel-2 imagery spanning 2013–2024. Validation against high-resolution unmanned aerial vehicle imagery yielded root mean square errors (RMSE) of 7.3–13.4 m and discrete Fréchet distances of 23.5–29.6 m, confirming robust algorithmic performance with decameter-level accuracy suitable for submesoscale front detection. Our analysis revealed the occurrence of pronounced spatiotemporal variability in front distributions: suspended sediment fronts exhibit distinct seasonal patterns with nearshore concentration (0–2 km) and peak occurrence around Nanhui Shoal, while plume fronts demonstrate greater offshore extension (2–5 km). Mechanistic analysis revealed that tidal forcing is the primary factor controlling the front dynamics, with suspended sediment fronts favoring neap tide conditions and plume fronts peaking during spring tide floods. These findings advance the fundamental understanding of estuarine front dynamics and provide quantitative frameworks for coastal environmental management and sustainable development strategies.
{"title":"Identification and characterization of estuarine submesoscale fronts based on multisource high-resolution remote sensing","authors":"Ruiyang Xu, Rui Yuan, Jing Chen","doi":"10.1016/j.jag.2026.105165","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105165","url":null,"abstract":"Estuarine fronts represent critical submesoscale oceanographic phenomena that govern material transport, pollutant dispersion, and ecosystem dynamics in coastal environments. In this study, we developed a comprehensive remote sensing framework for monitoring suspended sediment fronts and plume fronts in the Yangtze River Estuary–Hangzhou Bay transition zone, a region of exceptional ecological and economic significance. Leveraging the Google Earth Engine cloud computing platform, we developed an integrated multi-algorithm approach that combines spectral index enhancement, automated thresholding, clustering analysis, and edge detection techniques for high-precision front feature extraction from Landsat-8 and Sentinel-2 imagery spanning 2013–2024. Validation against high-resolution unmanned aerial vehicle imagery yielded root mean square errors (RMSE) of 7.3–13.4 m and discrete Fréchet distances of 23.5–29.6 m, confirming robust algorithmic performance with decameter-level accuracy suitable for submesoscale front detection. Our analysis revealed the occurrence of pronounced spatiotemporal variability in front distributions: suspended sediment fronts exhibit distinct seasonal patterns with nearshore concentration (0–2 km) and peak occurrence around Nanhui Shoal, while plume fronts demonstrate greater offshore extension (2–5 km). Mechanistic analysis revealed that tidal forcing is the primary factor controlling the front dynamics, with suspended sediment fronts favoring neap tide conditions and plume fronts peaking during spring tide floods. These findings advance the fundamental understanding of estuarine front dynamics and provide quantitative frameworks for coastal environmental management and sustainable development strategies.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"4 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209254","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Accurate delineation and mapping of forest burned areas (FBAs) are critical for assessing fire-induced ecological effects and informing post-fire management strategies. Although medium-resolution sensors (10–30 m) have advanced FBA detection, such detection at a finer spatial resolution remains challenging due to various confounding factors like cloud/terrain shadows and phenology-induced spectral confusions. To mitigate these false alarms, an accurate detection framework was developed, integrating active fire (AF) data with Sentinel-2 multi-temporal imagery. Potential candidates are identified using optimized spectral features across single- and dual-temporal scales. These candidates are subsequently refined by synergizing AF data and the novel smoothed Enhanced Burned Area Index (EBAI) imagery to enhance detection sensitivity. The EBAI leverages the temporal-spectral variations of FBAs, characterized by abrupt deviations from pre-fire reflectance and short-term post-fire anomalies, to highlight burned signals and suppress false alarms. Applied to the Amur River Basin, we generated the first 10-m resolution monthly FBA product (ARB_FBA) for 2020–2023. Validation with independent reference units demonstrated low commission errors (7.65–8.46%), low omission errors (9.43–13.02%), and high Dice coefficients (89.94–91.21%). ARB_FBA outperformed global products (GABAM, FireCCI51, and MCD64A1) by delivering more accurate, detailed, and extensive burned areas. Results indicated marked fluctuations in FBA, ranging from a minimum of 3.83 × 105 ha in 2021 to a peak of 1.08 × 106 ha in 2023. The 2023 surge was associated with record-breaking heatwaves and forest management deficiencies, while areal peaks confirmed spring (March to May) as the highest-risk period. These findings establish a robust foundation for post-fire impact assessments and offer actionable insights for forest management.
{"title":"Accurate forest burned area detection by integrating active fire data with enhanced Sentinel-2 multi-temporal imagery","authors":"Huixin Ren, Chunying Ren, Lin Li, Yeqiao Wang, Mingming Jia, Zongming Wang, Yanbiao Xi, Pan Liu, Chenzhen Xia, Qinglin Hou, Xing Ruan","doi":"10.1016/j.jag.2026.105163","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105163","url":null,"abstract":"Accurate delineation and mapping of forest burned areas (FBAs) are critical for assessing fire-induced ecological effects and informing post-fire management strategies. Although medium-resolution sensors (10–30 m) have advanced FBA detection, such detection at a finer spatial resolution remains challenging due to various confounding factors like cloud/terrain shadows and phenology-induced spectral confusions. To mitigate these false alarms, an accurate detection framework was developed, integrating active fire (AF) data with Sentinel-2 multi-temporal imagery. Potential candidates are identified using optimized spectral features across single- and dual-temporal scales. These candidates are subsequently refined by synergizing AF data and the novel smoothed Enhanced Burned Area Index (EBAI) imagery to enhance detection sensitivity. The EBAI leverages the temporal-spectral variations of FBAs, characterized by abrupt deviations from pre-fire reflectance and short-term post-fire anomalies, to highlight burned signals and suppress false alarms. Applied to the Amur River Basin, we generated the first 10-m resolution monthly FBA product (ARB_FBA) for 2020–2023. Validation with independent reference units demonstrated low commission errors (7.65–8.46%), low omission errors (9.43–13.02%), and high Dice coefficients (89.94–91.21%). ARB_FBA outperformed global products (GABAM, FireCCI51, and MCD64A1) by delivering more accurate, detailed, and extensive burned areas. Results indicated marked fluctuations in FBA, ranging from a minimum of 3.83 × 10<ce:sup loc=\"post\">5</ce:sup> ha in 2021 to a peak of 1.08 × 10<ce:sup loc=\"post\">6</ce:sup> ha in 2023. The 2023 surge was associated with record-breaking heatwaves and forest management deficiencies, while areal peaks confirmed spring (March to May) as the highest-risk period. These findings establish a robust foundation for post-fire impact assessments and offer actionable insights for forest management.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"321 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-12DOI: 10.1016/j.jag.2026.105159
Mathis Neuhauser, Alexandre Peltier, Thomas Ibanez, Marc Despinoy, Michel Le Page
Climate change is escalating the frequency and intensity of droughts and wildfires globally. In New Caledonia, future intensification is projected, yet the seasonal and directional temporal relationships between vegetation drought and wildfire activity remain insufficiently characterized at the regional scale. This study presents a regional case study combining remote sensing and in-situ datasets spanning 2000–2024, aggregated at monthly and municipal scales, to analyse drought–fire temporal interactions in New Caledonia. The approach quantifies the temporal sequencing of interactions between vegetation condition and burned area occurrence, distinguishing periods when vegetation stress precedes fire activity (fire hazard) from periods when fire occurrence is followed by altered vegetation conditions (post-fire vulnerability and potential feedbacks). Lagged and seasonal correlation analyses were conducted using the Vegetation Health Index (VHI) as a proxy of vegetation drought and the Burned areas Anomaly Index (BAI) to characterize wildfire activity. Results reveal robust and spatially coherent correlations between vegetation drought and burned area extent, highlighting the strong association between VHI and subsequent wildfire activity. Distinct seasonal interaction patterns emerge, with vegetation condition preceding fire activity during the early dry season (August–September), while fire occurrence is followed by modified surface conditions over subsequent months (November–May), consistent with post-fire ecosystem vulnerability and feedback mechanism. Marked geographic contrasts are observed, particularly between the west and east coasts. These findings improve understanding of drought–fire temporal linkages in New Caledonia and provide actionable, region-specific insights for seasonally targeted and spatially explicit wildfire risk management.
{"title":"Seasonal hazard-vulnerability patterns between drought and wildfire in New Caledonia derived from remote sensing products","authors":"Mathis Neuhauser, Alexandre Peltier, Thomas Ibanez, Marc Despinoy, Michel Le Page","doi":"10.1016/j.jag.2026.105159","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105159","url":null,"abstract":"Climate change is escalating the frequency and intensity of droughts and wildfires globally. In New Caledonia, future intensification is projected, yet the seasonal and directional temporal relationships between vegetation drought and wildfire activity remain insufficiently characterized at the regional scale. This study presents a regional case study combining remote sensing and in-situ datasets spanning 2000–2024, aggregated at monthly and municipal scales, to analyse drought–fire temporal interactions in New Caledonia. The approach quantifies the temporal sequencing of interactions between vegetation condition and burned area occurrence, distinguishing periods when vegetation stress precedes fire activity (fire hazard) from periods when fire occurrence is followed by altered vegetation conditions (post-fire vulnerability and potential feedbacks). Lagged and seasonal correlation analyses were conducted using the Vegetation Health Index (VHI) as a proxy of vegetation drought and the Burned areas Anomaly Index (BAI) to characterize wildfire activity. Results reveal robust and spatially coherent correlations between vegetation drought and burned area extent, highlighting the strong association between VHI and subsequent wildfire activity. Distinct seasonal interaction patterns emerge, with vegetation condition preceding fire activity during the early dry season (August–September), while fire occurrence is followed by modified surface conditions over subsequent months (November–May), consistent with post-fire ecosystem vulnerability and feedback mechanism. Marked geographic contrasts are observed, particularly between the west and east coasts. These findings improve understanding of drought–fire temporal linkages in New Caledonia and provide actionable, region-specific insights for seasonally targeted and spatially explicit wildfire risk management.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"1 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209288","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-12DOI: 10.1016/j.jag.2026.105153
Marta Sapena, Johannes Mast, Elisabeth Schoepfer, Hannes Taubenböck
Urban areas across Africa have undergone unprecedented growth, posing significant challenges for sustainable development, infrastructure planning, and climate resilience. Existing mapping products often struggle to capture the dynamic and heterogeneous nature of these evolving urban landscapes, highlighting the need for maps that are both up-to-date and locally relevant. This study introduces a practical, cloud-based solution: an online tool for site-specific mapping (i.e., tailored maps for a defined area of interest) that leverages the capabilities of Google Earth Engine. The tool uses Sentinel-1 and Sentinel-2 imagery to derive a wide range of spectral and texture metrics, supplemented by terrain data, and is trained using open building footprint datasets available for 2022. In an experimental setup, nine model configurations were tested under varying data availability conditions across 100 urban sites in Africa. The best-performing model achieved a mean F1-score of 0.59 (recall 0.63; precision 0.51) when validated against building footprints, with higher accuracy observed in dense urban areas. This configuration was integrated into the freely available ‘Do-it-yourself built-up mapping tool’ (DIY-BU). A quantitative analysis across the 100 test sites showed that the maps generated by our tool for 2022 were substantially more accurate (with an increase of F1-score by 0.18–0.30) than global multi-temporal products analysed for the same period (i.e., Dynamic World, ESRI land cover, GISA, GLC_FCS30D and GISD30). While the quantitative assessment was limited to the 2022 reference year, and the multi-temporal maps rely on a monotonic growth assumption (preventing the detection of demolition), a qualitative analysis highlighted the tool’s advantages in capturing detailed urban expansion and small-scale structures. The DIY-BU-mapping tool offers a valuable resource for a variety of applications, including urban planning, infrastructure monitoring, disaster preparedness and climate adaptation. Beyond presenting the tool’s functionality, the paper discusses its limitations and potential applications across diverse geographic and data availability contexts.
{"title":"Do-it-yourself built-up mapping tool: A practical cloud-based solution using Sentinel imagery for mapping urban expansion in Africa","authors":"Marta Sapena, Johannes Mast, Elisabeth Schoepfer, Hannes Taubenböck","doi":"10.1016/j.jag.2026.105153","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105153","url":null,"abstract":"Urban areas across Africa have undergone unprecedented growth, posing significant challenges for sustainable development, infrastructure planning, and climate resilience. Existing mapping products often struggle to capture the dynamic and heterogeneous nature of these evolving urban landscapes, highlighting the need for maps that are both up-to-date and locally relevant. This study introduces a practical, cloud-based solution: an online tool for site-specific mapping (i.e., tailored maps for a defined area of interest) that leverages the capabilities of Google Earth Engine. The tool uses Sentinel-1 and Sentinel-2 imagery to derive a wide range of spectral and texture metrics, supplemented by terrain data, and is trained using open building footprint datasets available for 2022. In an experimental setup, nine model configurations were tested under varying data availability conditions across 100 urban sites in Africa. The best-performing model achieved a mean F1-score of 0.59 (recall 0.63; precision 0.51) when validated against building footprints, with higher accuracy observed in dense urban areas. This configuration was integrated into the freely available ‘Do-it-yourself built-up mapping tool’ (DIY-BU). A quantitative analysis across the 100 test sites showed that the maps generated by our tool for 2022 were substantially more accurate (with an increase of F1-score by 0.18–0.30) than global multi-temporal products analysed for the same period (i.e., Dynamic World, ESRI land cover, GISA, GLC_FCS30D and GISD30). While the quantitative assessment was limited to the 2022 reference year, and the multi-temporal maps rely on a monotonic growth assumption (preventing the detection of demolition), a qualitative analysis highlighted the tool’s advantages in capturing detailed urban expansion and small-scale structures. The DIY-BU-mapping tool offers a valuable resource for a variety of applications, including urban planning, infrastructure monitoring, disaster preparedness and climate adaptation. Beyond presenting the tool’s functionality, the paper discusses its limitations and potential applications across diverse geographic and data availability contexts.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"96 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209291","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-12DOI: 10.1016/j.jag.2026.105156
Yan Jia, Quan Liu, Hongjie He, Shuanggen Jin, Chunqiao Song, Kyle Gao, Zebiao Wu
Water level can be effectively monitored through satellite remote sensing. However, a persistent challenge lies in balancing spatial coverage and resolution. High-resolution data provide detailed information but suffer from limited spatial coverage, while low-resolution data are inadequate for accurate large-scale monitoring. In this paper, a super-resolution approach with machine learning is proposed to enhance the accuracy of water level monitoring as a solution to this problem. By leveraging Real-world Enhanced Super-Resolution Generative Adversarial Network (Real-ESRGAN) and a high-quality pretrained model, this study enhances the spatial resolution of Sentinel-1 SAR imagery, whose features are then used to develop a water level estimation model, enabling dynamic monitoring of river water levels. Results show that after super-resolution, the Root Mean Squared Error (RMSE) is reduced to 0.322 m (∼30.6% decrease), the Nash-Sutcliffe Efficiency (NSE) increases to 0.957, and the Pearson Correlation Coefficient (R) improves to 0.987. The proposed super-resolution approach improves both spatial resolution (to 2.5 m/pixel) and water level estimation accuracy. Moreover, the results from representative stations, including densely urbanized regions, agricultural irrigation zones, and areas near sluices, show that our super-resolution approach provides accurate water level estimates and reliable trend predictions. Our experiments confirm its usefulness across a diverse spectrum of high accuracy hydrological monitoring scenarios.
{"title":"High accuracy water level estimation using super-resolution Sentinel-1 data","authors":"Yan Jia, Quan Liu, Hongjie He, Shuanggen Jin, Chunqiao Song, Kyle Gao, Zebiao Wu","doi":"10.1016/j.jag.2026.105156","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105156","url":null,"abstract":"Water level can be effectively monitored through satellite remote sensing. However, a persistent challenge lies in balancing spatial coverage and resolution. High-resolution data provide detailed information but suffer from limited spatial coverage, while low-resolution data are inadequate for accurate large-scale monitoring. In this paper, a super-resolution approach with machine learning is proposed to enhance the accuracy of water level monitoring as a solution to this problem. By leveraging Real-world Enhanced Super-Resolution Generative Adversarial Network (Real-ESRGAN) and a high-quality pretrained model, this study enhances the spatial resolution of Sentinel-1 SAR imagery, whose features are then used to develop a water level estimation model, enabling dynamic monitoring of river water levels. Results show that after super-resolution, the Root Mean Squared Error (RMSE) is reduced to 0.322 m (∼30.6% decrease), the Nash-Sutcliffe Efficiency (NSE) increases to 0.957, and the Pearson Correlation Coefficient (R) improves to 0.987. The proposed super-resolution approach improves both spatial resolution (to 2.5 m/pixel) and water level estimation accuracy. Moreover, the results from representative stations, including densely urbanized regions, agricultural irrigation zones, and areas near sluices, show that our super-resolution approach provides accurate water level estimates and reliable trend predictions. Our experiments confirm its usefulness across a diverse spectrum of high accuracy hydrological monitoring scenarios.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"59 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209290","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-02-12DOI: 10.1016/j.jag.2026.105162
Samuel McGowan, Jonathan Procter, Gabor Kereszturi, Fabien Albino, Indranova Suhendro
This study characterizes the amplitude changes associated with the 4 December 2021 eruption of Mount Semeru and maps the resulting deposits using Sentinel-1 SAR backscatter and PlanetScope optical imagery. Results show that lahar deposits caused surface smoothening which reduced SAR backscatter, whereas pyroclastic density currents (PDCs) increased backscatter due to higher moisture content. We have also shown the potential of differential polarimetric responses between surface cover types to identify areas of channel widening and vegetation destruction, providing a rapid means of identifying impacted areas in the context of volcanic crisis management. The supervised classification of SAR and high-resolution optical images enabled the production of an accurate geomorphological map able to separate different pyroclastic flow and lahar flow deposits both sedimentologically and spatially. Classified channelized lahar deposits were also used to quantify channel widening associated with the eruption, which significantly impacted the Supiturang Village.
{"title":"Analysis of the 4 December 2021 lahar on Mount Semeru using optical and SAR remote sensing","authors":"Samuel McGowan, Jonathan Procter, Gabor Kereszturi, Fabien Albino, Indranova Suhendro","doi":"10.1016/j.jag.2026.105162","DOIUrl":"https://doi.org/10.1016/j.jag.2026.105162","url":null,"abstract":"This study characterizes the amplitude changes associated with the 4 December 2021 eruption of Mount Semeru and maps the resulting deposits using Sentinel-1 SAR backscatter and PlanetScope optical imagery. Results show that lahar deposits caused surface smoothening which reduced SAR backscatter, whereas pyroclastic density currents (PDCs) increased backscatter due to higher moisture content. We have also shown the potential of differential polarimetric responses between surface cover types to identify areas of channel widening and vegetation destruction, providing a rapid means of identifying impacted areas in the context of volcanic crisis management. The supervised classification of SAR and high-resolution optical images enabled the production of an accurate geomorphological map able to separate different pyroclastic flow and lahar flow deposits both sedimentologically and spatially. Classified channelized lahar deposits were also used to quantify channel widening associated with the eruption, which significantly impacted the Supiturang Village.","PeriodicalId":50341,"journal":{"name":"International Journal of Applied Earth Observation and Geoinformation","volume":"106 1","pages":""},"PeriodicalIF":7.5,"publicationDate":"2026-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146209287","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：