Remote Sensing of Environment最新文献

{"title":"AgriFM: A multi-source temporal remote sensing foundation model for Agriculture mapping","authors":"Wenyuan Li ,&nbsp;Shunlin Liang ,&nbsp;Keyan Chen ,&nbsp;Yongzhe Chen ,&nbsp;Han Ma ,&nbsp;Jianglei Xu ,&nbsp;Yichuan Ma ,&nbsp;Yuxiang Zhang ,&nbsp;Shikang Guan ,&nbsp;Husheng Fang ,&nbsp;Zhenwei Shi","doi":"10.1016/j.rse.2026.115234","DOIUrl":"10.1016/j.rse.2026.115234","url":null,"abstract":"<div><div>Climate change and population growth intensify the demand for precise agriculture mapping to enhance food security. Such mapping tasks require robust modeling of multi-scale spatiotemporal patterns from fine field textures to landscape context, and from short-term phenology to full growing-season dynamics. Existing methods often process spatial and temporal features separately, limiting their ability to capture essential agricultural dynamics. While transformer-based remote sensing foundation models (RSFMs) offer unified spatiotemporal modeling ability, most of them remain suboptimal: they either use fixed windows that ignore multi-scale crop characteristics or neglect temporal information entirely. To address these gaps, we propose AgriFM, a multi-source, multi-temporal foundation model for agriculture mapping. AgriFM introduces a synchronized spatiotemporal downsampling strategy within a Video Swin Transformer backbone, enabling efficient handling of long and variable-length satellite time series while preserving multi-scale spatial and phenological information. It is pre-trained on a globally representative dataset comprising over 25 million samples from MODIS, Landsat-8/9, and Sentinel-2 with land cover fractions as pre-training supervision. AgriFM further integrates a versatile decoder specifically designed to dynamically fuse multi-source features from different stages of backbone and accommodate varying temporal lengths, thereby supporting consistent and scalable agriculture mapping across diverse satellite sources and task requirements. It supports diverse tasks including agricultural land mapping, field boundary delineation, agricultural land use/land cover mapping, and specific crop mapping (e.g., winter wheat and paddy rice) with different data sources. Comprehensive evaluations show that AgriFM consistently outperforms existing deep learning models and general-purpose RSFMs across multiple agriculture mapping tasks. Codes and models are available at <span><span>https://github.com/flyakon/AgriFM</span><svg><path></path></svg></span> and <span><span>https://glass.hku.hk</span><svg><path></path></svg></span></div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115234"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938802","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}

{"title":"Mapping the structural diversity of Central African and Western US forests using GEDI","authors":"Fabian D. Schneider ,&nbsp;Morgan Dean ,&nbsp;Elsa M. Ordway ,&nbsp;Moses B. Libalah ,&nbsp;Antonio Ferraz","doi":"10.1016/j.rse.2025.115215","DOIUrl":"10.1016/j.rse.2025.115215","url":null,"abstract":"<div><div>This study maps forest structural diversity, a key component of ecosystem diversity, using NASA's GEDI spaceborne lidar, providing new opportunities to support conservation and restoration efforts. Focusing on biodiversity hotspots in Central Africa and the Western US, we evaluated GEDI's ability to capture spatial variation in forest canopy structure by comparing GEDI-derived metrics with 391 km² of airborne laser scanning (ALS) data. Forest structural traits were assessed at 1 km² resolution, with GEDI showing robust correlations with ALS, particularly in dense and flat Central African forests (r² up to 0.85), and moderate agreement in more heterogeneous terrain in the California Sierra Nevada (r² up to 0.55). We quantified structural diversity as horizontal variation in canopy structure using a probability density-based multivariate diversity framework. GEDI canopy height (rh98), canopy cover, and foliage height diversity jointly captured independent axes of canopy variation and explained structural diversity derived from wall-to-wall ALS (r² = 0.37 at 1 km²). Resulting maps reveal high structural diversity in mid-elevation and coastal forests of the Western US and in forest-savanna transitions and volcanic ranges of Central Africa, consistent with gradients in disturbance, topography and aridity. Despite limitations related to sampling density, waveform noise and terrain complexity, this study demonstrates that GEDI's footprint-level metrics can be used directly to quantify forest structural diversity. These findings highlight the potential of spaceborne lidar to provide scalable, trait-based structural diversity indicators, while emphasizing that interpretation remains context dependent and benefits from integration with ecological data, environmental drivers and disturbance regimes.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115215"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145938804","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}

{"title":"Highest quality remote sensing reflectance database compiled from 20+ years of MODIS-aqua measurements","authors":"Longteng Zhao,&nbsp;Zhongping Lee,&nbsp;Xiaolong Yu,&nbsp;Tianhao Wang,&nbsp;Daosheng Wang,&nbsp;Shaoling Shang","doi":"10.1016/j.rse.2025.115195","DOIUrl":"10.1016/j.rse.2025.115195","url":null,"abstract":"<div><div>Remote sensing reflectance (<em>R</em>_<em>rs</em>) is a fundamental property in satellite ocean color remote sensing, which is critical for retrieving optical-biogeochemical properties and data-driven atmospheric correction algorithms. In this study, with three criteria applicable to ∼91% of the global ocean, we compiled a database of the highest quality <em>R</em>_<em>rs</em> (HQ_MODISA-<em>R</em>_<em>rs</em>) of oceanic waters based on 20+ years of ocean color measurements by the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Aqua satellite. While removing a large number of daily “standard” data products, our evaluation showed that the criteria for the highest-quality <em>R</em>_<em>rs</em> (CHQR) improved MODIS <em>R</em>_<em>rs</em> data consistency with benchmark <em>in situ R</em>_<em>rs</em> datasets, such as those from MOBY and AERONET-OC. After applying CHQR, analysis of imagery products in the South Pacific Ocean revealed that the coefficient of variation (CV) of <em>R</em>_<em>rs</em> among pixels reduced from 0.042 (standard quality control) to 0.030, along with enhanced temporal consistency, which indicates that this approach effectively filters abnormal data products. While such a dataset played a key role in the development of the cross-satellite atmospheric correction algorithm (Lee et al., 2024), we here further demonstrate that applications of HQ_MODISA-<em>R</em>_<em>rs</em> have ∼21.0% of oceanic areas between 50°S and 50°N showing reversed long-term trends of <em>R</em>_<em>rs</em> compared to the trend based on the standard <em>R</em>_<em>rs</em> product. We anticipate that this highest-quality <em>R</em>_<em>rs</em> database would not only improve our evaluation and understanding of long-term changes in various <em>R</em>_<em>rs</em>-derivative bio-optical properties of the global ocean, but also help to obtain consistent products among various satellite ocean color missions.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115195"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145731548","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}

{"title":"An integrated atmospheric-topographic correction framework for land surface reflectance estimation using a spatial-spectral Attention U-Net model","authors":"Yichuan Ma ,&nbsp;Shunlin Liang ,&nbsp;Han Ma ,&nbsp;Tao He ,&nbsp;Xiran Shi ,&nbsp;Wenyuan Li ,&nbsp;Dejun Cai ,&nbsp;Xiongxin Xiao ,&nbsp;Shikang Guan ,&nbsp;Weiwei Liu ,&nbsp;Jianglei Xu ,&nbsp;Yongzhe Chen ,&nbsp;Yuxiang Zhang","doi":"10.1016/j.rse.2025.115188","DOIUrl":"10.1016/j.rse.2025.115188","url":null,"abstract":"<div><div>Surface reflectance, a fundamental parameter for deriving high-level satellite products, is typically obtained through atmospheric correction. However, in rugged terrains, topographic effects substantially distort surface reflectance estimates. Many post-topographic correction methods for surface reflectance were proposed and applied to mitigate topographic effects. However, the coupled atmospheric and topographic influences in radiative transfer processes cause significant errors when these corrections are performed separately. For example, we observed over 50 % of pixels in the official Landsat 8 surface reflectance data across a complex terrain exhibited physically implausible negative values, predominantly in shadowed areas. Moreover, traditional pixel-by-pixel methods failed to leverage valuable spatial information for estimation. To address these limitations, we developed an integrated atmospheric and topographic correction framework, Unet-TopoFlat, leveraging a spatial-spectral Attention U-Net algorithm and a novel pseudo-topographic synthetic strategy. The pseudo-topographic synthetic strategy generated sufficient and robust topographically incorporated TOA radiance samples across different terrains, surfaces, and atmospheric conditions, using surface reflectance over flat terrains based on radiative transfer models, atmospheric parameters, random forest regression, and a mountainous radiative transfer parameterization scheme. Using Landsat 8 data as a proxy for evaluation, the Unet-TopoFlat was trained on 47,398 samples (256 × 256 pixels), leveraging multiple datasets. The Unet-TopoFlat model effectively captured the spatial and spectral relationships between TOA radiance and surface reflectance, achieving a relative root mean square error (rRMSE) of 4.5 %–6.2 % across 20,314 samples spanning different terrain, temporal, and spectral bands. Compared to the baseline Unet-FLAT model, which lacked topographic consideration and exhibited substantial uncertainties, Unet-TopoFlat effectively reduced topographic effects, lowering negative reflectance ratios from 55.5 % to 2.8 % while accurately recovering surface information and preserving spectral information. Moreover, the leaf area index (LAI) and snow cover mapping using our estimated surface reflectance were superior to those using official products, and deviations reached up to 2.4 for LAI and 8 % for snow cover mapping at the regional scale. Our proposed framework is not sensor-specific and can be potentially applied to multiple optical remotely sensed data.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115188"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145823488","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}

{"title":"Quantification of phytoplankton primary production from space: A revisit based on high-frequency observations with the aid of Himawari-8/AHI","authors":"Zhaoxin Li ,&nbsp;Wei Yang ,&nbsp;Huangrong Chen ,&nbsp;Chong Shi ,&nbsp;Husi Letu ,&nbsp;Fang Shen","doi":"10.1016/j.rse.2026.115238","DOIUrl":"10.1016/j.rse.2026.115238","url":null,"abstract":"<div><div>Synoptic quantification of phytoplankton depth-integrated primary production (IPP) has advanced significantly over recent decades by leveraging satellite observations and sophisticated IPP models. However, monthly mean upstream products from polar-orbiting satellites, <em>e.g.</em>, the Moderate Resolution Imaging Spectroradiometer (MODIS), are commonly used to generate IPP products, raising a concern about whether neglecting diurnal or daily IPP variabilities may compromise the accuracy of monthly- and annual-scale quantifications. Here, we aim to investigate this concern by comparing IPP quantified using high-frequency data at multiple timescales. A theoretical time-resolved model (TPM) was utilized for IPP modeling, driven by either diurnal photosynthetically available radiation (PAR) from the Advanced Himawari Imager (AHI) onboard Himawari-8 (H8) or daily PAR from MODIS. Preliminary evaluation against <em>in situ</em> measurements corroborated the superiority of AHI-based daily IPP estimation over MODIS, attributed to the robustness of AHI PAR data across varying sky conditions. Satellite IPP products were generated in the full-disk area of H8 between 2016 and 2019 under “daily-to-monthly-to-annual” (DtA) and “monthly-to-annual” (MtA) scenarios for comparison, using other requisite daily and gap-free biogeochemical products. Our analysis unveiled moderate spatiotemporal discrepancies between DtA-based IPP products from AHI and MODIS, confirming an overestimation in MODIS-derived monthly (&lt; 8%) and annual total IPP (∼5%). In contrast, under the MtA scenario, MODIS substantially overestimated monthly (∼14–30%) and annual total IPP (∼20%) and gave biased temporal trends (∼1.3–1.6 times higher) compared to DtA-based IPP estimates of AHI. The discrepancies between IPP products were largely subject to the cloud-induced variabilities in daily PAR products and ocean color data coverage. By upscaling our results to the global ocean, it is anticipated that the annual total IPP previously estimated from MODIS with TPM-like models is overestimated by at least 19%. This study emphasizes the necessity of modeling IPP at finer timescales using high-frequency observations and provides insights for improving IPP quantification with the aid of geostationary satellites.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115238"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145962692","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}

{"title":"Sentinel-1 imagery for wide-scale quantitative landslide vulnerability assessment of buildings","authors":"Francesco Poggi ,&nbsp;Francesco Caleca ,&nbsp;Olga Nardini ,&nbsp;Francesco Barbadori ,&nbsp;Matteo Del Soldato ,&nbsp;Claudio De Luca ,&nbsp;Francesco Casu ,&nbsp;Manuela Bonano ,&nbsp;Riccardo Lanari ,&nbsp;Veronica Tofani ,&nbsp;Federico Raspini","doi":"10.1016/j.rse.2025.115199","DOIUrl":"10.1016/j.rse.2025.115199","url":null,"abstract":"<div><div>The occurrence of landslides can potentially cause considerable economic impact on a global scale, resulting in damage to exposed structures and infrastructures, including buildings. In order to determine the most efficacious risk mitigation strategies, the scientific community is engaged in the analyses aimed at assessing the expected consequences of landslide activation/reactivation. This approach involves the implementation of quantitative risk assessment procedures, based on the definition of the landslide susceptibility and intensity, the identification of exposed elements, and the vulnerability assessment, which represents the most challenging parameter. The present work proposes an approach to the quantitative vulnerability assessment for buildings exposed to slow-kinematic landslides via empirical fragility and vulnerability curves, which express the probabilistic relationship between the damage severity to exposed buildings and the intensity of the landslide. A comprehensive database was compiled, collating information on landslide-induced damage to over four thousand buildings in the Northern Apennines of Central Italy. The landslide intensity is evaluated by exploiting the worldwide freely accessible Sentinel-1 SAR images, which have been processed by using the Parallel-Small BAseline (P-SBAS) – Differential SAR Interferometry (DInSAR) processing chain. The Institute for Electromagnetic Sensing of the Environment (IREA) of the National Research Council (CNR) of Italy is responsible for the processing of the Sentinel-1 SAR images, within the framework of an agreement with the Italian Ministry of Environment and Energy Security (MASE) for the detection and analysis of ground displacement over the entire Italian territory. The ultimate objective of the present study is the exploitation of the derived vulnerability curve into the quantitative risk assessment procedure for a comprehensive evaluation of buildings in the Northern Apennines.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115199"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145785692","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}

{"title":"Integrating classifier transfer and sample transfer strategies for in-season crop mapping based on sample weighting technique","authors":"Yunze Zang ,&nbsp;Junxiong Zhou ,&nbsp;Xuehong Chen ,&nbsp;Tianyu Liu ,&nbsp;Miaogen Shen ,&nbsp;Wei Yang ,&nbsp;Xiufang Zhu ,&nbsp;Fei Zhang ,&nbsp;Jin Chen","doi":"10.1016/j.rse.2025.115208","DOIUrl":"10.1016/j.rse.2025.115208","url":null,"abstract":"<div><div>Timely crop mapping is crucial for field management, policy formulation, phenological monitoring, and yield forecasting. However, acquiring sufficient labeled samples in the current year presents a formidable challenge for in-season mapping. Previously proposed solutions mainly include classifier transfer and sample transfer strategies. The classifier transfer strategy trains classifiers with historical samples associated with historical-year features and then transfers the trained historical sample classifiers (HSC) to classify remote sensing data in the current year; the sample transfer strategy generates trusted samples associated with current-year remote sensing features by predicting labels of the current-year sample based on some prior knowledge (e.g., crop rotation pattern) and then trains trusted sample classifiers (TSC) for current-year classification. However, the performance of the classifier-transfer strategy may degrade when there is large interannual feature variation, while the performance of the sample-transfer strategy depends on the reliability of the generated trusted samples. This study proposes a novel approach that integrates the above two strategies for in-season mapping through a sample weighting technique. Firstly, two sample sets, trusted samples and classified samples associated with current-year features, are generated by crop rotation prediction and HSC, respectively. Subsequently, based on an independent assumption between the rotational prediction errors and the current-year remote sensing features, the optimal weights of these two sample sets are derived based on the Bayesian principle. Finally, an optimal weighted sample classifier (OWSC) is trained using the weighted samples for in-season classification. To illustrate the robustness of the proposed OWSC, we compared it with different methods combined with various classification models across four regions with different interannual feature variation and crop rotation stability. Results demonstrated that OWSC maintained its advantages across various regions and different available lengths of historical crop-type sequences. Owing to its independence from specific classifiers, the proposed sample weighting method can be seamlessly applied to any classification model and thus continues to benefit from advances in classification algorithms. Additionally, sensitivity experiments regarding the uncertainty in trusted samples and historical crop-type sequences showed that OWSC performed stably across different scenarios. Therefore, OWSC provides a promising solution for in-season crop mapping without current-year samples.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115208"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145836274","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}

{"title":"AnyTime-CD: Self-supervised change detection in arbitrary-length dense time series of high-resolution remote sensing images","authors":"Yang Qu ,&nbsp;JiaYi Li ,&nbsp;Xin Huang","doi":"10.1016/j.rse.2026.115239","DOIUrl":"10.1016/j.rse.2026.115239","url":null,"abstract":"<div><div>With the continuous enhancement of the observation capabilities of remote sensing satellites, dense multi-temporal high-resolution remote sensing images have become an important data source for monitoring surface changes. However, multi-temporal change detection (MTCD) remains challenging due to inconsistent time intervals, spatial misalignments, and a scarcity of labeled data, all of which hinder the generalization and practical deployment of deep learning models in large-scale applications. To address the above challenges, this paper proposed a multi-task self-supervised change detection framework (i.e., AnyTime-CD), which supports the input of time series of any length and can jointly model temporal-spectral-spatial features without manual annotation. This framework consists of three sub-tasks: temporal dynamic modeling, spectral semantic consistency learning, and spatial feature alignment, which respectively address the issues of irregular time interval, spectral perturbation, and spatial offset, thereby collaboratively optimize network performance. We evaluated AnyTime-CD on two globally distributed multi-temporal high-resolution datasets. The experimental results showed that, under an unsupervised setting, this method achieved performance close to or even surpassing that of the supervised methods. Compared with the best-performing self-supervised method, it achieved relative gains of over 13% (spatial) and 8% (temporal) in F1 scores (using a one-frame tolerance strategy), respectively. After sample fine-tuning, it further achieved F1 scores of 86.40% (spatial) and 79.50% (temporal), exceeding the state-of-the-art supervised method by 4.44% and 9.09%, respectively. Furthermore, AnyTime-CD showed good adaptability to different pre-training configurations. It is worth noting that it exhibited strong robustness under medium and low cloud cover conditions and could even utilize cloud perturbations as data augmentation to further enhance performance. In conclusion, AnyTime-CD offers a flexible and label-free solution for MTCD tasks, suitable for complex, dynamic, and interference-prone remote sensing scenarios, and is expected to promote the application of self-supervised methods in surface monitoring.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115239"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145962691","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}

{"title":"Began+: Leveraging bi-temporal SAR-optical data fusion to reconstruct clear-sky satellite imagery under large cloud cover","authors":"Yu Xia ,&nbsp;Wei He ,&nbsp;Liangpei Zhang ,&nbsp;Hongyan Zhang","doi":"10.1016/j.rse.2025.115171","DOIUrl":"10.1016/j.rse.2025.115171","url":null,"abstract":"<div><div>In recent years, optical remote sensing imagery has played an increasingly vital role in Earth observation, but cloud contamination exists as an inevitable degradation. Combining synthetic aperture radar (SAR) and optical data with machine learning offers a promising solution for reconstructing clear-sky satellite imagery. Nevertheless, several challenges persist, including insufficient attention to large cloud cover, difficulties in restoring temporal changes, and limited practicality of deep models. To address these issues, this paper introduces a novel deep learning-based cloud removal framework, termed Began+, which integrates bi-temporal SAR-optical data to deal with cloudy images with high cover ratios. The Began+ framework comprises two primary components: a deep network and a flexible post-processing step, combining the strengths of data-driven models for restoring change information and traditional gap-filling algorithms for mitigating radiance discrepancies. First, a bi-output enhanced generative adversarial network, abbreviated as Began, is designed for image synthesis, featuring an enhanced channel-wise fusion block (ECFB) and a multi-scale depth-wise convolution residual block (MDRB). By applying the dual-tasking optimization and co-learning strategy, the Began model identifies potential change areas from bi-temporal SAR and pre-temporal optical inputs, guiding the synthesis of target optical images. Second, a range of cloud masking and gap-filling techniques can be optionally employed to effectively reduce radiometric discrepancies between the synthesized images and the cloudy data, ultimately yielding high-quality, clear-sky imagery. To meet the big data requirements of deep learning, we constructed two globally distributed cloud removal datasets, named BiS1L8-CR and BiS1S2-CR. Supported by these datasets, extensive experiments demonstrated that the Began+ framework effectively captures bi-temporal change features, reconstructing precise surface information in both Landsat-8 and Sentinel-2 satellite images under large cloud cover. Compared to the latest solutions and algorithms, our proposed Began+ framework exhibits significant advantages from both qualitative and quantitative perspectives in both simulated and real experiments. Furthermore, without strict constraints on input timing, the Began+ framework enables accurate reconstruction of large-scale dual-sensor imagery under high-ratio cloud cover, effectively restoring changing surfaces and improving the quality of unsupervised vegetation extraction.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115171"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145732131","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}

{"title":"Mapping wide-area land subsidence from groundwater use in the North China plain by machine learning-based InSAR adjustment","authors":"Mi Jiang ,&nbsp;Zhou Wu ,&nbsp;Xudong Wang ,&nbsp;Lin Bai ,&nbsp;Zhiwei Li ,&nbsp;Zhong Lu","doi":"10.1016/j.rse.2025.115226","DOIUrl":"10.1016/j.rse.2025.115226","url":null,"abstract":"<div><div>Since 2014, a series of China's water resource management policies have been implemented to mitigate groundwater-induced land subsidence in the North China Plain (NCP). While previous studies have demonstrated the benefits of Synthetic Aperture Radar interferometry (InSAR) in providing policy-relevant insights into the spatio-temporal dynamics of subsidence and groundwater recovery, most have focused on localized regions, leaving the long-term impact of these measures across the entire NCP insufficiently evaluated. A key challenge is the variation of long-wavelength errors in each SAR frame, which results in inconsistencies in the subsidence velocity field over large areas. To address this issue, this paper proposes a machine learning-based adjustment approach for routinely wide-area subsidence mapping and then fully evaluating land subsidence and associated groundwater depletion in the NCP from the end of 2014 to 2022. The novelty of this method lies in the adaptive selection of the optimal model for each SAR frame, which minimizes the varying long-wavelength errors, rather than relying on a unified model for all SAR frames as commonly used in state-of-the-art approaches. Additionally, we mitigated the difference of InSAR measurement in the overlap regions between consecutive tracks caused by the varying incidence angles by incorporating GNSS data and a plate motion model. Using synthetic and real Sentinel-1 data, we validated the performance of the proposed method against prevalent approaches through an independent GNSS validation dataset, demonstrating accuracy improvements from 3.8-17.5 mm/yr to 2.0 mm/yr. The results indicated that approximately 56,882 km² of the NCP area experienced land subsidence greater than 20 mm/yr. The central alluvial and coastal plains were the primary areas of subsidence, with a maximum cumulative subsidence of up to 2 m. The average subsidence velocity peaked in 2018 at 38.5 mm/yr. Subsidence has been alleviated after 2021. Our results revealed the lower bound of groundwater loss from the confined aquifer in the NCP, totaling 24.9 billion m³ between the end of 2014 and 2022. Of this total, 20.2 billion m³ (81 %) was lost from October 2014 to the end of 2020, with the loss decreasing to 4.7 billion m³ (19 %) during the period from January 2021 to December 2022. This study provides new evidence for China's groundwater management practices in addressing land subsidence in the NCP.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"334 ","pages":"Article 115226"},"PeriodicalIF":11.4,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145902873","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}