Pub Date : 2026-01-09DOI: 10.1016/j.srs.2025.100356
Jianhong Wu , Xiaojuan Liu , Longshan Yang , Yongze Song , Hong Cai , Bowen Yu
Hyperspectral unmixing (HU) is an essential HSI processing technique for separating the spectral signatures of endmembers and estimating their corresponding abundances within mixed pixels. While nonnegative matrix factorization (NMF) has been widely used for HU with a single-layer structure, it cannot fully explore the hidden features in a HSI. In this paper, an improved optimization deep nonnegative matrix factorization (IODNMF) method is proposed, which improves the parameter optimization approach of the deep NMF model for exploring the deep feature representation of HSI to achieve unsupervised HU. With both unknown endmembers and abundances, a heuristic algorithm is utilized to initialize the parameters of each layer to speed up the convergence of the proposed method. To ensure the nonnegativity of parameters and avoid gradient vanishing and explosion during training, a multiplicative update rule based on positive–negative separation is devised to update the endmembers and abundances in the pre-training and fine-tuning stages. In addition, two layer-setting strategies and three per-layer parameter setting strategies are proposed to effectively solve the network structure setting problem in deep NMF methods. Experimental results on synthetic and real HSI datasets show that the proposed deep NMF algorithm performs more effectively than other classical unmixing methods and achieves a significant improvement in unmixing performance.
{"title":"An improved optimization deep nonnegative matrix factorization for hyperspectral unmixing","authors":"Jianhong Wu , Xiaojuan Liu , Longshan Yang , Yongze Song , Hong Cai , Bowen Yu","doi":"10.1016/j.srs.2025.100356","DOIUrl":"10.1016/j.srs.2025.100356","url":null,"abstract":"<div><div>Hyperspectral unmixing (HU) is an essential HSI processing technique for separating the spectral signatures of endmembers and estimating their corresponding abundances within mixed pixels. While nonnegative matrix factorization (NMF) has been widely used for HU with a single-layer structure, it cannot fully explore the hidden features in a HSI. In this paper, an improved optimization deep nonnegative matrix factorization (IODNMF) method is proposed, which improves the parameter optimization approach of the deep NMF model for exploring the deep feature representation of HSI to achieve unsupervised HU. With both unknown endmembers and abundances, a heuristic algorithm is utilized to initialize the parameters of each layer to speed up the convergence of the proposed method. To ensure the nonnegativity of parameters and avoid gradient vanishing and explosion during training, a multiplicative update rule based on positive–negative separation is devised to update the endmembers and abundances in the pre-training and fine-tuning stages. In addition, two layer-setting strategies and three per-layer parameter setting strategies are proposed to effectively solve the network structure setting problem in deep NMF methods. Experimental results on synthetic and real HSI datasets show that the proposed deep NMF algorithm performs more effectively than other classical unmixing methods and achieves a significant improvement in unmixing performance.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100356"},"PeriodicalIF":5.2,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976983","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}
Pub Date : 2026-01-09DOI: 10.1016/j.srs.2025.100361
Sihan Xue , Haibin Shi , Xianyue Li , Jianwen Yan , Weigang Wang , Qingfeng Miao , Yan Yan , Cong Hou , Yi Zhao , Xinlu Li
Persistent cloud contamination severely affects optical satellite-based crop classification, creating critical gaps in phenological monitoring essential for food security. This study presents GS-3D FCN, a novel framework integrating Gap Filling-Savitzky-Golay (GF-SG) time series reconstruction with 3D Fully Convolutional Networks (3D FCN) for robust crop mapping under extreme cloud conditions. The method reconstructs cloud-contaminated pixels using GF-SG algorithm and leverages MODIS-Landsat fusion to generate continuous NDVI time series. These reconstructed layers are integrated as spectral-temporal dimensions into the 3D FCN architecture, enabling hierarchical extraction of crop phenological patterns. An enhanced Automatic Cloud Cover Assessment algorithm adaptively adjusts decision criteria to maximize valid pixel retention. To address severe class imbalance from cloud occlusion, we apply Synthetic Minority Over-sampling Technique with Gaussian Noise (SMOGN) combined with Cross-entropy loss Middle Supervision (CE-MidS) and Supervised Contrastive loss Middle Supervision (SupCon-MidS) intermediate supervision strategies, facilitating discriminative feature learning. Applied to the cloud-prone South Bank Yellow River irrigation area in Inner Mongolia, GS-3D FCN achieves 93.80 % overall accuracy across four sites. The framework demonstrates exceptional stability, maintaining 85.2 % accuracy at 80 % cloud coverage compared to 33.6–37.2 % accuracy degradation in traditional methods. The GF-SG component achieves RMSE of 0.0853 with edge preservation index of 0.0085. Analysis reveals the critical role of dataset balance in crop mapping under cloud interference. The proposed GS-3D FCN framework provides a simple and effective solution for crop mapping in cloud-prone areas. By integrating near-real-time NDVI data sources, the framework provides a feasible path for accurate crop monitoring in global continuous cloudy areas, and provides a new perspective for future research on the interaction between extreme climate and crop mapping.
{"title":"Crop classification mapping in multi-cloud Regions: An integrated approach using GF-SG time series reconstruction and 3D deep convolutional networks","authors":"Sihan Xue , Haibin Shi , Xianyue Li , Jianwen Yan , Weigang Wang , Qingfeng Miao , Yan Yan , Cong Hou , Yi Zhao , Xinlu Li","doi":"10.1016/j.srs.2025.100361","DOIUrl":"10.1016/j.srs.2025.100361","url":null,"abstract":"<div><div>Persistent cloud contamination severely affects optical satellite-based crop classification, creating critical gaps in phenological monitoring essential for food security. This study presents GS-3D FCN, a novel framework integrating Gap Filling-Savitzky-Golay (GF-SG) time series reconstruction with 3D Fully Convolutional Networks (3D FCN) for robust crop mapping under extreme cloud conditions. The method reconstructs cloud-contaminated pixels using GF-SG algorithm and leverages MODIS-Landsat fusion to generate continuous NDVI time series. These reconstructed layers are integrated as spectral-temporal dimensions into the 3D FCN architecture, enabling hierarchical extraction of crop phenological patterns. An enhanced Automatic Cloud Cover Assessment algorithm adaptively adjusts decision criteria to maximize valid pixel retention. To address severe class imbalance from cloud occlusion, we apply Synthetic Minority Over-sampling Technique with Gaussian Noise (SMOGN) combined with Cross-entropy loss Middle Supervision (CE-MidS) and Supervised Contrastive loss Middle Supervision (SupCon-MidS) intermediate supervision strategies, facilitating discriminative feature learning. Applied to the cloud-prone South Bank Yellow River irrigation area in Inner Mongolia, GS-3D FCN achieves 93.80 % overall accuracy across four sites. The framework demonstrates exceptional stability, maintaining 85.2 % accuracy at 80 % cloud coverage compared to 33.6–37.2 % accuracy degradation in traditional methods. The GF-SG component achieves RMSE of 0.0853 with edge preservation index of 0.0085. Analysis reveals the critical role of dataset balance in crop mapping under cloud interference. The proposed GS-3D FCN framework provides a simple and effective solution for crop mapping in cloud-prone areas. By integrating near-real-time NDVI data sources, the framework provides a feasible path for accurate crop monitoring in global continuous cloudy areas, and provides a new perspective for future research on the interaction between extreme climate and crop mapping.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100361"},"PeriodicalIF":5.2,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037868","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}
Pub Date : 2026-01-08DOI: 10.1016/j.srs.2026.100367
Juan Quiros-Vargas , Cosimo Brogi , Alexander Damm , Bastian Siegmann , Patrick Rademske , Vicente Burchard-Levine , Vera Krieger , Marius Schmidt , Jan Hanuš , Mauricio Martello , Lutz Weihermüller , Onno Muller , Uwe Rascher
Restrictions in the soil water availability can strongly impact crop productivity. The increasing frequency and severity of drought events, as a result of global warming, has made the assessment of drought stress effects on vegetation of utmost importance for meeting humanity's agricultural production needs. Recent advances in remote sensing of solar-induced chlorophyll fluorescence (SIF) provide a basis for new approaches to directly assess crop water status, since SIF is closely related to photosynthesis and, thus, to early plant physiological processes triggered by limitations in the water supply. This study provides new insights into the effect of varying levels of plant available water (PAW) in the soil on SIF emissions. We used several SIF datasets acquired with the high-performance airborne imaging spectrometer HyPlant during five subsequent vegetation periods (2018, 2019, 2020, 2021 and 2022), each having a different precipitation regime. We normalized the SIF maps for the underlying effects of canopy structure, calculated SIF emission efficiency (eSIF) and selected various crop fields including sugar beet, wheat and potato. Maps of eSIF were compared with spatial PAW patterns, which were derived from a forward soil infiltration model. Our results show positive correlation between eSIF and PAW in rainfed sugar beet fields at early growing stage, which remained consistent when accounting for variations in the leaf area index (LAI). This suggests that eSIF variations in sugar beet reflect the spatial reduction of photosynthesis caused by reduced PAW. In irrigated potato fields, conversely, no eSIF-PAW correlations were found. This indicates the absence of leaf-level water stress in these well-irrigated fields. In rainfed winter wheat fields that were already in a late developmental stage, the variations in the SIF signal were dominated by locally different ripening, i.e., chlorophyll degradation, and therefore not representative of changing PAW. With this study, we could demonstrate that normalized airborne SIF measurements are related to the functional water stress response in different crops. This study supports future investigations on the development of SIF-based tools for the improvement of water management in agriculture.
{"title":"Solar-induced chlorophyll fluorescence (SIF) tracks variations in the soil-plant available water (PAW): a multiyear analysis on three crops","authors":"Juan Quiros-Vargas , Cosimo Brogi , Alexander Damm , Bastian Siegmann , Patrick Rademske , Vicente Burchard-Levine , Vera Krieger , Marius Schmidt , Jan Hanuš , Mauricio Martello , Lutz Weihermüller , Onno Muller , Uwe Rascher","doi":"10.1016/j.srs.2026.100367","DOIUrl":"10.1016/j.srs.2026.100367","url":null,"abstract":"<div><div>Restrictions in the soil water availability can strongly impact crop productivity. The increasing frequency and severity of drought events, as a result of global warming, has made the assessment of drought stress effects on vegetation of utmost importance for meeting humanity's agricultural production needs. Recent advances in remote sensing of solar-induced chlorophyll fluorescence (SIF) provide a basis for new approaches to directly assess crop water status, since SIF is closely related to photosynthesis and, thus, to early plant physiological processes triggered by limitations in the water supply. This study provides new insights into the effect of varying levels of plant available water (PAW) in the soil on SIF emissions. We used several SIF datasets acquired with the high-performance airborne imaging spectrometer HyPlant during five subsequent vegetation periods (2018, 2019, 2020, 2021 and 2022), each having a different precipitation regime. We normalized the SIF maps for the underlying effects of canopy structure, calculated SIF emission efficiency (eSIF) and selected various crop fields including sugar beet, wheat and potato. Maps of eSIF were compared with spatial PAW patterns, which were derived from a forward soil infiltration model. Our results show positive correlation between eSIF and PAW in rainfed sugar beet fields at early growing stage, which remained consistent when accounting for variations in the leaf area index (LAI). This suggests that eSIF variations in sugar beet reflect the spatial reduction of photosynthesis caused by reduced PAW. In irrigated potato fields, conversely, no eSIF-PAW correlations were found. This indicates the absence of leaf-level water stress in these well-irrigated fields. In rainfed winter wheat fields that were already in a late developmental stage, the variations in the SIF signal were dominated by locally different ripening, i.e., chlorophyll degradation, and therefore not representative of changing PAW. With this study, we could demonstrate that normalized airborne SIF measurements are related to the functional water stress response in different crops. This study supports future investigations on the development of SIF-based tools for the improvement of water management in agriculture.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100367"},"PeriodicalIF":5.2,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977575","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}
Pub Date : 2026-01-07DOI: 10.1016/j.srs.2026.100363
Tian Xia, Yanan Zhao, Liguang Jiang
Satellite altimetry has been increasingly used in monitoring inland water bodies. Waveform retracking plays a major role in water level retrieval. However, there remain many challenges to retrieving accurate river water levels, especially for rivers surrounded by various water bodies. In this study, we investigated this problem by diagnosing six retrackers in the Yangtze River, where the environment is very complex. Results show that the official retracker (i.e., OCOG and threshold retrackers) used in Sentinel-3 product exhibits varying performance across 12 virtual stations, with RMSE in the range of 0.55–2.76 m. Surprisingly, no one retracker performs consistently well across all virtual stations. The enhanced multiple waveform persistent peak (MWaPP+) retracker was slightly better than the others. Taking multiple waveforms into consideration is a better strategy than single waveform-based ones. Poor performance is due to irregular waveforms, which are attributed to various water bodies surrounding the river. The number, elevation, and proportion of anomalous water bodies within the footprint are found decisive. In such complex environments, a combination of multiple strategies is needed to improve the accuracy of retrieved water levels. The proposed strategy, by combining FFSAR and MWaPP+, substantially enhanced accuracy and the number of observations. Nevertheless, we call for a round robin exercise to test more retracking strategies to deal with this problem.
{"title":"Altimetry river water level retrieval over complex environments: assessment and diagnosis of different strategies","authors":"Tian Xia, Yanan Zhao, Liguang Jiang","doi":"10.1016/j.srs.2026.100363","DOIUrl":"10.1016/j.srs.2026.100363","url":null,"abstract":"<div><div>Satellite altimetry has been increasingly used in monitoring inland water bodies. Waveform retracking plays a major role in water level retrieval. However, there remain many challenges to retrieving accurate river water levels, especially for rivers surrounded by various water bodies. In this study, we investigated this problem by diagnosing six retrackers in the Yangtze River, where the environment is very complex. Results show that the official retracker (i.e., OCOG and threshold retrackers) used in Sentinel-3 product exhibits varying performance across 12 virtual stations, with RMSE in the range of 0.55–2.76 m. Surprisingly, no one retracker performs consistently well across all virtual stations. The enhanced multiple waveform persistent peak (MWaPP+) retracker was slightly better than the others. Taking multiple waveforms into consideration is a better strategy than single waveform-based ones. Poor performance is due to irregular waveforms, which are attributed to various water bodies surrounding the river. The number, elevation, and proportion of anomalous water bodies within the footprint are found decisive. In such complex environments, a combination of multiple strategies is needed to improve the accuracy of retrieved water levels. The proposed strategy, by combining FFSAR and MWaPP+, substantially enhanced accuracy and the number of observations. Nevertheless, we call for a round robin exercise to test more retracking strategies to deal with this problem.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100363"},"PeriodicalIF":5.2,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939365","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}
Pub Date : 2026-01-07DOI: 10.1016/j.srs.2026.100369
Aiping Jiang , Dongsheng Wang , Tiantian Jin , Peng Li , Tao Xu , Di Zhang , Boran Zhu , Junqiang Lin , Qidong Peng
Reservoir construction has led to frequent algal blooms in tributaries of backwater areas, threatening water quality, aquatic ecological safety, and public health. The mechanisms governing algal blooms in reservoir bays are complex, leading to spatiotemporal occurrence patterns that are difficult to accurately identify. In this study, Sentinel-2 imagery from 2019 to 2023 were used to characterize the spatiotemporal distribution of algal blooms in Xiangxi Bay (XXB) near the dam of the Three Gorges Reservoir (TGR). The Google Earth Engine (GEE) platform was used for remote sensing image selection, preprocessing, and spectral index calculation, while random forest (RF) model and polynomial regression were employed to develop chlorophyll-a (Chl-a) retrieval model. Box plots and the Mann–Whitney U test were used to compare bloom and non-bloom conditions across nine hydrometeorological variables from the perspective of external influences, while the internal bloom mechanisms were analyzed by integrating the critical depth theory of algal growth with the hydrodynamic mixing characteristics and bloom occurrence patterns in XXB. The results indicated that: (1) Sentinel-2 imagery effectively captures bloom dynamics in narrow tributary embayments, and the Normalized Difference Chlorophyll Index (NDCI) accurately retrieves Chl-a, with the retrieval model achieving an R2 of 0.76; (2) In recent years, algal blooms in XXB have been more likely to occur in March, July, August, and September, particularly in August, with approximately 18 % of the XXB experiencing blooms, predominantly at level II intensity. High-frequency bloom areas are mainly distributed in the mid-to upper reaches of the backwater zone; (3) During the flood season, mid-layer intrusion from the Yangtze River and upstream tributary inflow promote stable stratification in XXB, especially under low-rainfall conditions, thereby favoring algal bloom formation in the mid-to upper reaches. Enhancing hydrodynamic disturbance through physical mixing or reservoir inflow and water-level regulation, particularly under low 10-day rainfall (<35 mm), can effectively suppress bloom development.
{"title":"Remote sensing facilitates the exploration of algal bloom dynamics and its hydrometeorological drivers in tributary bays of the Three Gorges Reservoir","authors":"Aiping Jiang , Dongsheng Wang , Tiantian Jin , Peng Li , Tao Xu , Di Zhang , Boran Zhu , Junqiang Lin , Qidong Peng","doi":"10.1016/j.srs.2026.100369","DOIUrl":"10.1016/j.srs.2026.100369","url":null,"abstract":"<div><div>Reservoir construction has led to frequent algal blooms in tributaries of backwater areas, threatening water quality, aquatic ecological safety, and public health. The mechanisms governing algal blooms in reservoir bays are complex, leading to spatiotemporal occurrence patterns that are difficult to accurately identify. In this study, Sentinel-2 imagery from 2019 to 2023 were used to characterize the spatiotemporal distribution of algal blooms in Xiangxi Bay (XXB) near the dam of the Three Gorges Reservoir (TGR). The Google Earth Engine (GEE) platform was used for remote sensing image selection, preprocessing, and spectral index calculation, while random forest (RF) model and polynomial regression were employed to develop chlorophyll-a (Chl-a) retrieval model. Box plots and the Mann–Whitney <em>U</em> test were used to compare bloom and non-bloom conditions across nine hydrometeorological variables from the perspective of external influences, while the internal bloom mechanisms were analyzed by integrating the critical depth theory of algal growth with the hydrodynamic mixing characteristics and bloom occurrence patterns in XXB. The results indicated that: (1) Sentinel-2 imagery effectively captures bloom dynamics in narrow tributary embayments, and the Normalized Difference Chlorophyll Index (NDCI) accurately retrieves Chl-a, with the retrieval model achieving an R<sup>2</sup> of 0.76; (2) In recent years, algal blooms in XXB have been more likely to occur in March, July, August, and September, particularly in August, with approximately 18 % of the XXB experiencing blooms, predominantly at level II intensity. High-frequency bloom areas are mainly distributed in the mid-to upper reaches of the backwater zone; (3) During the flood season, mid-layer intrusion from the Yangtze River and upstream tributary inflow promote stable stratification in XXB, especially under low-rainfall conditions, thereby favoring algal bloom formation in the mid-to upper reaches. Enhancing hydrodynamic disturbance through physical mixing or reservoir inflow and water-level regulation, particularly under low 10-day rainfall (<35 mm), can effectively suppress bloom development.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100369"},"PeriodicalIF":5.2,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146037964","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}
Pub Date : 2026-01-06DOI: 10.1016/j.srs.2026.100365
E. Korchemkina , T. Churilova , E. Skorokhod , N. Moiseeva , T. Efimova
A data set including in situ absorption of particulate and dissolved colored organic matter, absorption of phytoplankton and chlorophyll-a concentration was measured during the research cruise on board R/V “Professor Multanovsky” on August–September 2023 in shelf waters of the south Kamchatka Peninsula. In order to adjust the existing semi-analytical algorithm to the optically complex waters, the regional parameterization of in-water optically active components absorption was used. Based on in situ bio-optical properties the reflectance spectra were modeled and used for the algorithm testing. The algorithm allows to separate the absorption of phytoplankton and colored detrital matter using the separate spectral sites for their calculation. The total non-water absorption was retrieved with high accuracy, while phytoplankton absorption (chlorophyll-a concentration) was overestimated on average by 35 %, and CDM absorption was underestimated on average by 32 %. Accuracy of their retrievals allows the application of the algorithm for ecological monitoring. An analysis showed that the reflectance spectral shape is strictly determined by the total non-water absorption. The high variability of ratios between optically active components leads to weak connection between reflectance spectral shape and specific optical components. Further refining of the semi-analytical algorithm consists in a selection of more suitable spectral sites and switching between them based on the contributions of components to the total absorption.
{"title":"Development of the satellite bio-optical algorithm for the shelf waters along the southern Kamchatka Peninsula: effect of optically active components variability on the spectral remote sensing reflectance","authors":"E. Korchemkina , T. Churilova , E. Skorokhod , N. Moiseeva , T. Efimova","doi":"10.1016/j.srs.2026.100365","DOIUrl":"10.1016/j.srs.2026.100365","url":null,"abstract":"<div><div>A data set including <em>in situ</em> absorption of particulate and dissolved colored organic matter, absorption of phytoplankton and chlorophyll-a concentration was measured during the research cruise on board R/V “Professor Multanovsky” on August–September 2023 in shelf waters of the south Kamchatka Peninsula. In order to adjust the existing semi-analytical algorithm to the optically complex waters, the regional parameterization of in-water optically active components absorption was used. Based on <em>in situ</em> bio-optical properties the reflectance spectra were modeled and used for the algorithm testing. The algorithm allows to separate the absorption of phytoplankton and colored detrital matter using the separate spectral sites for their calculation. The total non-water absorption was retrieved with high accuracy, while phytoplankton absorption (chlorophyll-a concentration) was overestimated on average by 35 %, and CDM absorption was underestimated on average by 32 %. Accuracy of their retrievals allows the application of the algorithm for ecological monitoring. An analysis showed that the reflectance spectral shape is strictly determined by the total non-water absorption. The high variability of ratios between optically active components leads to weak connection between reflectance spectral shape and specific optical components. Further refining of the semi-analytical algorithm consists in a selection of more suitable spectral sites and switching between them based on the contributions of components to the total absorption.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100365"},"PeriodicalIF":5.2,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977576","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}
Pub Date : 2026-01-06DOI: 10.1016/j.srs.2026.100364
Fangjie Li , Inbal Becker-Reshef , Josef Wagner , Françoise Nerry
Timely and accurate winter wheat mapping is essential for agricultural monitoring and food security. However, efficiently acquiring high-quality training data for supervised classification remains a challenge. In this study, we developed a rule-based method to automatically generate training samples using Sentinel-2 green chlorophyll vegetation index (GCVI) time series. Then, the key phenological periods were identified through feature importance analysis, and spectral features from these periods were used with a Random Forest (RF) classifier to produce 10 m resolution winter wheat distribution maps for Hengshui, Kaifeng, and Xiangyang in 2022 and 2023. To evaluate temporal transferability, the automatically generated training samples from 2022 to 2023 were transferred to subsequent years, enabling winter wheat mapping for 2023 and 2024 based on cross-year training data. Accuracy assessments showed that the proposed method achieved high performance, with average overall accuracy (OA) of 96.04 ± 1.97 % and 94.81 ± 2.14 % in 2022 and 2023, respectively, and average F1 scores of 91.21 % and 90.83 %. The winter wheat maps generated using transferred samples also demonstrated good temporal transferability, and maintained high accuracy, with average OA of 94.06 ± 2.19 % in 2023 and 94.58 ± 2.01 % in 2024. Area estimates from stratified random sampling showed that winter wheat planting areas in Hengshui, Kaifeng, and Xiangyang were 290.8 ± 16.82, 199.16 ± 10.65, and 318.05 ± 44.16 thousand hectares (kha) in 2022, increasing to 379.34 ± 19.75, 209.27 ± 12.68, and 342.02 ± 42.81 kha in 2023, respectively. Compared with existing winter wheat products, the map generated in this study achieved higher classification accuracy and finer spatial detail. Overall, this study provides a practical and effective approach for automatic training sample generation in winter wheat mapping, and offers valuable guidance for large-scale, long-term agricultural monitoring.
{"title":"Rule-based training sample generation using Sentinel-2 GCVI time series for winter wheat mapping","authors":"Fangjie Li , Inbal Becker-Reshef , Josef Wagner , Françoise Nerry","doi":"10.1016/j.srs.2026.100364","DOIUrl":"10.1016/j.srs.2026.100364","url":null,"abstract":"<div><div>Timely and accurate winter wheat mapping is essential for agricultural monitoring and food security. However, efficiently acquiring high-quality training data for supervised classification remains a challenge. In this study, we developed a rule-based method to automatically generate training samples using Sentinel-2 green chlorophyll vegetation index (GCVI) time series. Then, the key phenological periods were identified through feature importance analysis, and spectral features from these periods were used with a Random Forest (RF) classifier to produce 10 m resolution winter wheat distribution maps for Hengshui, Kaifeng, and Xiangyang in 2022 and 2023. To evaluate temporal transferability, the automatically generated training samples from 2022 to 2023 were transferred to subsequent years, enabling winter wheat mapping for 2023 and 2024 based on cross-year training data. Accuracy assessments showed that the proposed method achieved high performance, with average overall accuracy (OA) of 96.04 ± 1.97 % and 94.81 ± 2.14 % in 2022 and 2023, respectively, and average F1 scores of 91.21 % and 90.83 %. The winter wheat maps generated using transferred samples also demonstrated good temporal transferability, and maintained high accuracy, with average OA of 94.06 ± 2.19 % in 2023 and 94.58 ± 2.01 % in 2024. Area estimates from stratified random sampling showed that winter wheat planting areas in Hengshui, Kaifeng, and Xiangyang were 290.8 ± 16.82, 199.16 ± 10.65, and 318.05 ± 44.16 thousand hectares (kha) in 2022, increasing to 379.34 ± 19.75, 209.27 ± 12.68, and 342.02 ± 42.81 kha in 2023, respectively. Compared with existing winter wheat products, the map generated in this study achieved higher classification accuracy and finer spatial detail. Overall, this study provides a practical and effective approach for automatic training sample generation in winter wheat mapping, and offers valuable guidance for large-scale, long-term agricultural monitoring.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100364"},"PeriodicalIF":5.2,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939336","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}
Pub Date : 2026-01-05DOI: 10.1016/j.srs.2026.100366
Weidong Xu , Martin J. Wooster , Jiangping He , Andrea Meraner , Jose Gomez-Dans , Zixia Liu , Isabel F. Trigo , Emanuel Dutra
Geostationary Earth Observation satellites, originally developed for weather forecasting, offer unique high temporal resolution imaging capabilities increasingly suited for detecting the fast-changing dynamics of landscape fires. The newly operational Meteosat Third Generation (MTG) satellite carries a Flexible Combined Imager (FCI) that greatly improves on the spatial, temporal and radiometric characteristics of the predecessor Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) system. Here we describe for the first time the application of an active fire (AF) detection algorithm to FCI data, and the retrieval of fire radiative power (FRP) estimates from the detected AF pixels. The algorithm used is the Fire Thermal Anomaly (FTA) approach, currently used to generate the operational SEVIRI AF data products at the EUMETSAT Land Surface Analysis Satellite Application Facility (LSA SAF). A comparative analysis between the FCI-derived outputs and those obtained from the existing SEVIRI system is undertaken in order to evaluate the benefits provided by FCI. We also include in this comparison data products from the polar-orbiting Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) systems. Our intercomparisons made in detail over specific Portuguese and Greek wildfires, and systematically across Africa and Europe, reveals four key findings: (1) FCI detects fire onset up to 4 h earlier than SEVIRI for the specific fires examined, and 2 h before MODIS and 4 h before VIIRS; (2) FCI generated 5 × more AF pixel detections than SEVIRI, due to a much reduced minimum FRP detection threshold (∼10 MW versus ∼40 MW) enabling the detection of the many AF pixels missed by SEVIRI; (3) FCI AF detection errors of omission were 38 % compared to MODIS centre-of-scan data, and 68 % compared to VIIRS, substantially improving on SEVIRI's 83 % and 89 % respectively; while commission errors compared to these two remained low at 12 % and 10 % respectively; (4) FCI FRP retrievals showed very strong agreement with the matching ones provided by MODIS (r2 = 0.97, slope = 0.93). FCI offers detections every 10 min over the full disk, and 2.5 min over Europe when rapid-scan commences after launch of the second MTG Imagery platform. The results shown here suggest that the operational active fire data products based on FCI and planned to be issued from the EUMETSAT LSA SAF using the FTA algorithm should deliver a substantial improvement in satellite-based fire monitoring across Europe and Africa compared to the already successful products currently generated from MSG.
{"title":"Major improvements in spaceborne early fire detection and small-fire FRP retrieval with the meteosat third generation flexible combined imager","authors":"Weidong Xu , Martin J. Wooster , Jiangping He , Andrea Meraner , Jose Gomez-Dans , Zixia Liu , Isabel F. Trigo , Emanuel Dutra","doi":"10.1016/j.srs.2026.100366","DOIUrl":"10.1016/j.srs.2026.100366","url":null,"abstract":"<div><div>Geostationary Earth Observation satellites, originally developed for weather forecasting, offer unique high temporal resolution imaging capabilities increasingly suited for detecting the fast-changing dynamics of landscape fires. The newly operational Meteosat Third Generation (MTG) satellite carries a Flexible Combined Imager (FCI) that greatly improves on the spatial, temporal and radiometric characteristics of the predecessor Meteosat Second Generation (MSG) Spinning Enhanced Visible and Infrared Imager (SEVIRI) system. Here we describe for the first time the application of an active fire (AF) detection algorithm to FCI data, and the retrieval of fire radiative power (FRP) estimates from the detected AF pixels. The algorithm used is the Fire Thermal Anomaly (FTA) approach, currently used to generate the operational SEVIRI AF data products at the EUMETSAT Land Surface Analysis Satellite Application Facility (LSA SAF). A comparative analysis between the FCI-derived outputs and those obtained from the existing SEVIRI system is undertaken in order to evaluate the benefits provided by FCI. We also include in this comparison data products from the polar-orbiting Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) systems. Our intercomparisons made in detail over specific Portuguese and Greek wildfires, and systematically across Africa and Europe, reveals four key findings: (1) FCI detects fire onset up to 4 h earlier than SEVIRI for the specific fires examined, and 2 h before MODIS and 4 h before VIIRS; (2) FCI generated 5 × more AF pixel detections than SEVIRI, due to a much reduced minimum FRP detection threshold (∼10 MW versus ∼40 MW) enabling the detection of the many AF pixels missed by SEVIRI; (3) FCI AF detection errors of omission were 38 % compared to MODIS centre-of-scan data, and 68 % compared to VIIRS, substantially improving on SEVIRI's 83 % and 89 % respectively; while commission errors compared to these two remained low at 12 % and 10 % respectively; (4) FCI FRP retrievals showed very strong agreement with the matching ones provided by MODIS (r<sup>2</sup> = 0.97, slope = 0.93). FCI offers detections every 10 min over the full disk, and 2.5 min over Europe when rapid-scan commences after launch of the second MTG Imagery platform. The results shown here suggest that the operational active fire data products based on FCI and planned to be issued from the EUMETSAT LSA SAF using the FTA algorithm should deliver a substantial improvement in satellite-based fire monitoring across Europe and Africa compared to the already successful products currently generated from MSG.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100366"},"PeriodicalIF":5.2,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976986","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}
Pub Date : 2026-01-05DOI: 10.1016/j.srs.2026.100368
Jianyong Cui , Wenwen Gao , Chunlei Meng
Surface roughness is a key parameter in meteorological simulations and wind energy assessments, and its spatial distribution is influenced by various factors. However, the complexity of these factors makes it difficult to retrieve roughness. Although machine learning methods have partially addressed this issue, they still face the challenge of insufficient measurement data. To tackle this, the present study proposes a knowledge distillation framework that integrates physical models and machine learning. It establishes a “teacher-student” model to enable knowledge transfer from regions with sufficient data to target regions with zero samples. In the source domain, where abundant ground truth data is available, four models-Random Forest, Support Vector Regression, Multi-layer Perceptron, and Transformer—were trained. The Multi-layer Perceptron, which achieved the best performance (correlation coefficient: 0.81, RMSE: 0.74, MAE: 0.51), was selected as the teacher model. Then, using the knowledge distillation method, soft labels were generated from remote sensing data in the target region to guide the training of the student model. This facilitated cross-domain knowledge transfer. The results show that the student model's training accuracy improved to 0.89, with the RMSE and MAE reduced to 0.62 and 0.33, respectively, significantly outperforming the teacher model. Compared to ERA5 reanalysis data and land surface model results, the student model's inversion of surface roughness in the target region reduced the mean absolute error by approximately 18 %, effectively solving the parameter estimation problem under the condition of no measurement samples. This study significantly enhances the accuracy of surface roughness estimation and provides more reliable parameter input for meteorological simulations and numerical weather forecasting.
{"title":"Calculation of surface roughness using machine learning algorithms combined with knowledge distillation","authors":"Jianyong Cui , Wenwen Gao , Chunlei Meng","doi":"10.1016/j.srs.2026.100368","DOIUrl":"10.1016/j.srs.2026.100368","url":null,"abstract":"<div><div>Surface roughness is a key parameter in meteorological simulations and wind energy assessments, and its spatial distribution is influenced by various factors. However, the complexity of these factors makes it difficult to retrieve roughness. Although machine learning methods have partially addressed this issue, they still face the challenge of insufficient measurement data. To tackle this, the present study proposes a knowledge distillation framework that integrates physical models and machine learning. It establishes a “teacher-student” model to enable knowledge transfer from regions with sufficient data to target regions with zero samples. In the source domain, where abundant ground truth data is available, four models-Random Forest, Support Vector Regression, Multi-layer Perceptron, and Transformer—were trained. The Multi-layer Perceptron, which achieved the best performance (correlation coefficient: 0.81, RMSE: 0.74, MAE: 0.51), was selected as the teacher model. Then, using the knowledge distillation method, soft labels were generated from remote sensing data in the target region to guide the training of the student model. This facilitated cross-domain knowledge transfer. The results show that the student model's training accuracy improved to 0.89, with the RMSE and MAE reduced to 0.62 and 0.33, respectively, significantly outperforming the teacher model. Compared to ERA5 reanalysis data and land surface model results, the student model's inversion of surface roughness in the target region reduced the mean absolute error by approximately 18 %, effectively solving the parameter estimation problem under the condition of no measurement samples. This study significantly enhances the accuracy of surface roughness estimation and provides more reliable parameter input for meteorological simulations and numerical weather forecasting.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100368"},"PeriodicalIF":5.2,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939366","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}
Pub Date : 2026-01-02DOI: 10.1016/j.srs.2025.100362
Tahir Sattar , Majid Nazeer , Man Sing Wong , Janet Elizabeth Nichol , Xiaolin Zhu
Mangroves are the resistant species found in the intertidal zones, providing ecosystem services such as protection of shorelines, provision of habitats to flora and fauna, and contributing to nutrient cycling. Study of their leaf properties has always been challenging, but this has been facilitated by the advent of Hyperspectral Imaging (HSI) systems. In such a context, this study undertook the development of a hyperspectral library offering the reflectance characteristics for adaxial and abaxial surfaces of mangrove species found in Hong Kong, on the temporal scale of seven days to facilitate the species identification and monitor the leaf decay. This library contained species level data, plot level data, and decay level data. Field surveys in fifteen plots (900 m2 each) conducted in the Eastern and Western regions of Hong Kong collected hyperspectral data of five mangrove species, namely: Ceriops tagal, Kandelia obovata, Avicennia marina, Avicennia germinans, and Aegiceras corniculatum, using two different types of HSI systems i.e., Specim IQ (in-field data) and NEO Hyspex (in-lab data) hyperspectral cameras. A comparison of sensors unveiled a notably higher reflectance in field collected data than that of the lab-collected data, with a range of 11.8 % (Kandelia obovate) to 73.1 % (Aegiceras corniculatum). The Root Mean Square Error (RMSE) indicated deviation between the two sensors, i.e., 0.211 for Ceriops tagal, followed by Kandelia obovata (0.233), Avicennia marina (0.317), Avicennia germinans, and Aegiceras corniculatum (0.349). This freely available comprehensive hyperspectral library will serve as the foundation for training datasets to achieve automated classification with enhanced accuracy. This open access hyperspectral library will assist the researchers to relate the physiological and anatomical variations in leaves with the changes in hyperspectral reflectance on the temporal scale.
{"title":"Establishing a hyperspectral library for Hong Kong mangroves: Species differentiation and leaf decay dynamics","authors":"Tahir Sattar , Majid Nazeer , Man Sing Wong , Janet Elizabeth Nichol , Xiaolin Zhu","doi":"10.1016/j.srs.2025.100362","DOIUrl":"10.1016/j.srs.2025.100362","url":null,"abstract":"<div><div>Mangroves are the resistant species found in the intertidal zones, providing ecosystem services such as protection of shorelines, provision of habitats to flora and fauna, and contributing to nutrient cycling. Study of their leaf properties has always been challenging, but this has been facilitated by the advent of Hyperspectral Imaging (HSI) systems. In such a context, this study undertook the development of a hyperspectral library offering the reflectance characteristics for adaxial and abaxial surfaces of mangrove species found in Hong Kong, on the temporal scale of seven days to facilitate the species identification and monitor the leaf decay. This library contained species level data, plot level data, and decay level data. Field surveys in fifteen plots (900 m<sup>2</sup> each) conducted in the Eastern and Western regions of Hong Kong collected hyperspectral data of five mangrove species, namely: <em>Ceriops tagal</em>, <em>Kandelia obovata, Avicennia marina, Avicennia germinans, and Aegiceras corniculatum,</em> using two different types of HSI systems i.e., Specim IQ (in-field data) and NEO Hyspex (in-lab data) hyperspectral cameras. A comparison of sensors unveiled a notably higher reflectance in field collected data than that of the lab-collected data, with a range of 11.8 % (Kandelia obovate) to 73.1 % (<em>Aegiceras corniculatum</em>). The Root Mean Square Error (RMSE) indicated deviation between the two sensors, i.e., 0.211 for <em>Ceriops tagal</em>, followed by <em>Kandelia obovata</em> (0.233), <em>Avicennia marina</em> (0.317), <em>Avicennia germinans</em>, and <em>Aegiceras corniculatum</em> (0.349). This freely available comprehensive hyperspectral library will serve as the foundation for training datasets to achieve automated classification with enhanced accuracy. This open access hyperspectral library will assist the researchers to relate the physiological and anatomical variations in leaves with the changes in hyperspectral reflectance on the temporal scale.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100362"},"PeriodicalIF":5.2,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938947","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号