Pub Date : 2026-01-17DOI: 10.1016/j.jag.2026.105117
Hui Liu , Wei Tong , Ning Chen , Tao Xie , Chenjia Huang , Xia Yue , Zhou Huang
Recent advances in deep learning and multimodal data fusion technologies have significantly enhanced hyperspectral image (HSI) classification performance. Nevertheless, classification accuracy of hyperspectral data continues to degrade substantially under diverse degradation scenarios, such as noise interference, spectral distortion, or reduced resolution. To robustly address this challenge, this paper proposes a novel cross-modal guided classification framework that integrates active remote sensing data (e.g., LiDAR) to improve classification resilience under degraded conditions. Specifically, we introduce a Cross-Modal Feature Pyramid Guidance (CMFPG) module, which effectively utilizes cross-modal information across multiple levels and scales to guide hyperspectral feature extraction and fusion, thereby enhancing modeling stability in degraded environments. Additionally, we develop the HyperGroupMix module, which enhances cross-domain adaptability through grouping spectral bands, extracting statistical features, and transferring features across samples. Experimental results conducted under complex degradation conditions demonstrate that our proposed method exhibits stable high-level classification accuracy and robustness in overall performance. The code is accessible at: https://github.com/miliwww/CMGF
{"title":"Seeing through the noise: A cross-modal guided framework for hyperspectral image classification under multi-type degradations","authors":"Hui Liu , Wei Tong , Ning Chen , Tao Xie , Chenjia Huang , Xia Yue , Zhou Huang","doi":"10.1016/j.jag.2026.105117","DOIUrl":"10.1016/j.jag.2026.105117","url":null,"abstract":"<div><div>Recent advances in deep learning and multimodal data fusion technologies have significantly enhanced hyperspectral image (HSI) classification performance. Nevertheless, classification accuracy of hyperspectral data continues to degrade substantially under diverse degradation scenarios, such as noise interference, spectral distortion, or reduced resolution. To robustly address this challenge, this paper proposes a novel cross-modal guided classification framework that integrates active remote sensing data (e.g., LiDAR) to improve classification resilience under degraded conditions. Specifically, we introduce a Cross-Modal Feature Pyramid Guidance (CMFPG) module, which effectively utilizes cross-modal information across multiple levels and scales to guide hyperspectral feature extraction and fusion, thereby enhancing modeling stability in degraded environments. Additionally, we develop the HyperGroupMix module, which enhances cross-domain adaptability through grouping spectral bands, extracting statistical features, and transferring features across samples. Experimental results conducted under complex degradation conditions demonstrate that our proposed method exhibits stable high-level classification accuracy and robustness in overall performance. The code is accessible at: <span><span>https://github.com/miliwww/CMGF</span><svg><path></path></svg></span></div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105117"},"PeriodicalIF":8.6,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976981","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 : 2026-01-17DOI: 10.1016/j.jag.2025.105081
Anjar Dimara Sakti , Tirto Prakoso , Cokro Santoso , Juan Andrean Milliandza , Pranda Mulya Putra Garniwa , Tri Muji Susantoro , Ketut Wikantika , Agung Budi Harto
Southeast Asia faces rapid growth in energy demand and continues to depend heavily on coal-based generation, creating an urgent need for renewable alternatives that can be deployed at scale. Biomass residues from agriculture represent an abundant but underutilized resource in the region. This study develops a machine learning–based spatial economic framework to quantify biomass energy potential from paddy, oil palm, cassava, and sugarcane residues across eight Southeast Asian countries and assess the feasibility of these residues for hybrid power generation. Crop yields were estimated using Random Forest regression with high-resolution (5 m) remote sensing predictors, achieving model performance achieving a maximum R2 of 0.628. Biomass residues were converted into electricity potential using crop-specific residue-to-product ratios and availability coefficients. The results show that Indonesia and Malaysia possess the highest agricultural residue potential from paddy and oil palm, while sugarcane residues exceed 20,000 MW across the region, with notable concentrations in Laos. Techno-economic modeling indicates that the levelized cost of electricity (LCOE) ranges from 0.04 to 0.11 USD/kWh, with payback times of 20–130 months, demonstrating cost competitiveness with coal, especially when the monetized cost of CO2 emissions are included. Spatial hybrid integration analysis reveals that paddy-rich corridors near existing coal plants have the strongest potential for biomass co-firing and hybridization. The proposed framework provides a scalable methodology for regional biomass planning and offers practical insights for policymakers in accelerating renewable energy transition and reducing fossil fuel dependence in Southeast Asia.
{"title":"Machine learning-based high resolution spatial economic modeling of biomass energy potential in Southeast Asia","authors":"Anjar Dimara Sakti , Tirto Prakoso , Cokro Santoso , Juan Andrean Milliandza , Pranda Mulya Putra Garniwa , Tri Muji Susantoro , Ketut Wikantika , Agung Budi Harto","doi":"10.1016/j.jag.2025.105081","DOIUrl":"10.1016/j.jag.2025.105081","url":null,"abstract":"<div><div>Southeast Asia faces rapid growth in energy demand and continues to depend heavily on coal-based generation, creating an urgent need for renewable alternatives that can be deployed at scale. Biomass residues from agriculture represent an abundant but underutilized resource in the region. This study develops a machine learning–based spatial economic framework to quantify biomass energy potential from paddy, oil palm, cassava, and sugarcane residues across eight Southeast Asian countries and assess the feasibility of these residues for hybrid power generation. Crop yields were estimated using Random Forest regression with high-resolution (5 m) remote sensing predictors, achieving model performance achieving a maximum R<sup>2</sup> of 0.628. Biomass residues were converted into electricity potential using crop-specific residue-to-product ratios and availability coefficients. The results show that Indonesia and Malaysia possess the highest agricultural residue potential from paddy and oil palm, while sugarcane residues exceed 20,000 MW across the region, with notable concentrations in Laos. Techno-economic modeling indicates that the levelized cost of electricity (LCOE) ranges from 0.04 to 0.11 USD/kWh, with payback times of 20–130 months, demonstrating cost competitiveness with coal, especially when the monetized cost of CO<sub>2</sub> emissions are included. Spatial hybrid integration analysis reveals that paddy-rich corridors near existing coal plants have the strongest potential for biomass co-firing and hybridization. The proposed framework provides a scalable methodology for regional biomass planning and offers practical insights for policymakers in accelerating renewable energy transition and reducing fossil fuel dependence in Southeast Asia.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105081"},"PeriodicalIF":8.6,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977786","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}
Building height is critical for understanding urban morphology and promoting sustainable growth. Although numerous approaches using Synthetic Aperture Radar (SAR) and optical images have been proposed for estimating building height, two key challenges remain: 1) neglecting edge characteristics results in inaccurate building boundary delineation; and 2) failure to capture both global and local context reduces height estimation reliability. To address these limitations, a novel edge-aware and locally-aggregated Mamba model is proposed, namely EL-Mamba. In this model, an edge-aware module is designed to enhance building boundary representation through multi-scale Laplacian of Gaussian (LoG) filtering. Additionally, a local aggregation strategy is integrated into Mamba’s global scanning mechanism, which enables the network to effectively capture global and local context. Building height data from Wuhan and the Yangtze River Delta (YRD) region in China is selected to evaluate EL-Mamba’s performance. The results indicate that compared to the four existing methods, EL-Mamba achieves the lowest root mean square error (RMSE), with values of 4.963 and 5.358 on the Wuhan and YRD datasets, respectively. Furthermore, EL-Mamba exhibits high computational efficiency and reliable generalization capability, indicating its significant potential for application in large-scale areas. Our implementation is available at: https://github.com/ohXu/EL_Mamba.
{"title":"EL-Mamba: An edge-aware and locally-aggregated Mamba network for building height estimation using Sentinel-1 and Sentinel-2 imagery","authors":"Qingyang Xu, Xuefeng Guan, Xu Li, Xiangyang Yang, Yifan Teng, Huayi Wu","doi":"10.1016/j.jag.2026.105103","DOIUrl":"10.1016/j.jag.2026.105103","url":null,"abstract":"<div><div>Building height is critical for understanding urban morphology and promoting sustainable growth. Although numerous approaches using Synthetic Aperture Radar (SAR) and optical images have been proposed for estimating building height, two key challenges remain: 1) neglecting edge characteristics results in inaccurate building boundary delineation; and 2) failure to capture both global and local context reduces height estimation reliability. To address these limitations, a novel edge-aware and locally-aggregated Mamba model is proposed, namely EL-Mamba. In this model, an edge-aware module is designed to enhance building boundary representation through multi-scale Laplacian of Gaussian (LoG) filtering. Additionally, a local aggregation strategy is integrated into Mamba’s global scanning mechanism, which enables the network to effectively capture global and local context. Building height data from Wuhan and the Yangtze River Delta (YRD) region in China is selected to evaluate EL-Mamba’s performance. The results indicate that compared to the four existing methods, EL-Mamba achieves the lowest root mean square error (RMSE), with values of 4.963 and 5.358 on the Wuhan and YRD datasets, respectively. Furthermore, EL-Mamba exhibits high computational efficiency and reliable generalization capability, indicating its significant potential for application in large-scale areas. Our implementation is available at: <span><span>https://github.com/ohXu/EL_Mamba</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105103"},"PeriodicalIF":8.6,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976982","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 : 2026-01-16DOI: 10.1016/j.jag.2026.105106
Xiran Zhou , Honghao Li , Zhenfeng Shao , Wenwen Li , Zhigang Yan
Maps are fundamental media for people to comprehend the spatial and temporal dimensions of a place. Currently, the majority of maps are generated through volunteered efforts, which significantly extends the critical and participatory aspects of geospatial information and provide diverse perspectives and scopes in terms of the geospatial characteristics of a place. However, most of these crowdsourced volunteered maps are often limited in spatial resolution due to file size, editing errors, printing limitations, and image quality issues, among others. This challenge poses a demand for deep learning-enhanced super-resolution reconstruction that can effectively convert low-resolution maps into high-resolution ones. Our previous research has revealed that map reconstruction presents unique requirements for preserving global content and local details involving map annotations and elements, which differ from the objectives of general image reconstruction. In this paper, we integrate MapSR—a CNN-based single map super-resolution reconstruction method we previously developed—into a GAN framework comprising a generator module, a discriminator module, and a local discriminant learning module. This integrated framework enables reconstructing both global map content and individual map elements. To testify the performance of CartoSR, we design three sets of experiments involving comparison with state-of-the-art methods for map super-resolution reconstruction, assessment of GANs’ effectiveness in map reconstruction, and analysis of CartoSR’s scalability. Experimental results demonstrate that our proposed CartoSR achieves state-of-the-art performance in single map super-resolution reconstruction. We hope it can serve as a routine for future research in this area.
{"title":"CartoSR: An attention-enhanced deep GANs for single map super resolution reconstruction","authors":"Xiran Zhou , Honghao Li , Zhenfeng Shao , Wenwen Li , Zhigang Yan","doi":"10.1016/j.jag.2026.105106","DOIUrl":"10.1016/j.jag.2026.105106","url":null,"abstract":"<div><div>Maps are fundamental media for people to comprehend the spatial and temporal dimensions of a place. Currently, the majority of maps are generated through volunteered efforts, which significantly extends the critical and participatory aspects of geospatial information and provide diverse perspectives and scopes in terms of the geospatial characteristics of a place. However, most of these crowdsourced volunteered maps are often limited in spatial resolution due to file size, editing errors, printing limitations, and image quality issues, among others. This challenge poses a demand for deep learning-enhanced super-resolution reconstruction that can effectively convert low-resolution maps into high-resolution ones. Our previous research has revealed that map reconstruction presents unique requirements for preserving global content and local details involving map annotations and elements, which differ from the objectives of general image reconstruction. In this paper, we integrate MapSR—a CNN-based single map super-resolution reconstruction method we previously developed—into a GAN framework comprising a generator module, a discriminator module, and a local discriminant learning module. This integrated framework enables reconstructing both global map content and individual map elements. To testify the performance of CartoSR, we design three sets of experiments involving comparison with state-of-the-art methods for map super-resolution reconstruction, assessment of GANs’ effectiveness in map reconstruction, and analysis of CartoSR’s scalability. Experimental results demonstrate that our proposed CartoSR achieves state-of-the-art performance in single map super-resolution reconstruction. We hope it can serve as a routine for future research in this area.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105106"},"PeriodicalIF":8.6,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977785","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 : 2026-01-16DOI: 10.1016/j.jag.2025.105084
Aru Han , Xorgan Uranghai , Li Mei , Youli Dong , Guoyue Yan , An Huang , Yuhai Bao , Song Qing , Azzaya Jukov , Urtnasan Mandakh , Tsambaa Battseren , Almaz Borjigidai
Heavy metal (HM) contamination of soil increasingly threatens the safety of Chinese herbal medicines. Hyperspectral remote sensing technology enables rapid, nondestructive monitoring of soil HM content. This study employs hyperspectral technology to indirectly monitor low Cr and Ni concentrations in the Soil-Scutellaria baicalensis(Huangqin)system using canopy spectra. Surveys and sampling, and data collection were conducted across five growth stages in two regions (Beijing and Hebei Province) of Scutellaria baicalensis (S. b.). The migration, correlation, and accumulation Cr and Ni in the Soil- S. b. system were analyzed. Canopy spectra were analyzed to estimate Cr and Ni contents using optimized spectral indices and LSSVM-based full-band modeling. Finally, Cr and Ni contents in S. b. leaves were estimated using the optimal inversion model, and further inversions based on enrichment characteristics were performed. The results were as follows. (1) In the May samples, the average soil Cr and Ni contents were 59.35 and 19.87 mg/kg, respectively, which were lower than the soil background values in Beijing and Hebei. Cr and Ni contents in roots, stems and leaves (RSL) were 26.96, 34.81, 37.09 mg/kg and 2.66, 4.08, 3.89 mg/kg, respectively. From June to September, the average Cr and Ni contents in the RSL of S. b. were 1.40, 0.76, 1.09 mg/kg and 1.11, 0.70, 1.23 mg/kg, respectively. (2) A strong correlation existed between low concentrations of HM in the soil and S. b. different parts, reaching a highly significant level (p < 0.01), supporting the use of S. b. leaves to estimate Cr and Ni in different soil parts and S. b.. (3) For Cr and Ni, by modeling the preprocessed first derivative (FD) and second derivative (SD) spectra, the Rc2 = 0.98, RMSEc = 0.10 mg/kg, Rv2 = 0.55, and RMSEv = 0.33 mg/kg for Cr, and Rc2 = 0.99, RMSEc = 0.04 mg/kg, Rv2 = 0.48, and RMSEv = 0.34 mg/kg for Ni were obtained, demonstrating their strong ability to estimate Cr and Ni in S. b. leaves. (4) Using hyperspectral estimation of Cr and Ni in S. b. leaves, together with grouped inversion of enrichment characteristics, stems and roots exceeded 0.33. Therefore, canopy spectral analysis combined with enrichment patterns offers a practical method for monitoring Cr and Ni in Huangqin system, supporting the safety testing of Chinese herbal medicines.
{"title":"Inversion of low heavy metal content in Soil-Scutellaria baicalensis systems using optimized spectral indices and LSSVM","authors":"Aru Han , Xorgan Uranghai , Li Mei , Youli Dong , Guoyue Yan , An Huang , Yuhai Bao , Song Qing , Azzaya Jukov , Urtnasan Mandakh , Tsambaa Battseren , Almaz Borjigidai","doi":"10.1016/j.jag.2025.105084","DOIUrl":"10.1016/j.jag.2025.105084","url":null,"abstract":"<div><div>Heavy metal (HM) contamination of soil increasingly threatens the safety of Chinese herbal medicines. Hyperspectral remote sensing technology enables rapid, nondestructive monitoring of soil HM content. This study employs hyperspectral technology to indirectly monitor low <em>Cr</em> and <em>Ni</em> concentrations in the Soil-<em>Scutellaria baicalensis</em>(Huangqin)system using canopy spectra. Surveys and sampling, and data collection were conducted across five growth stages in two regions (Beijing and Hebei Province) of <em>Scutellaria baicalensis (S. b.)</em>. The migration, correlation, and accumulation <em>Cr</em> and <em>Ni</em> in the Soil- <em>S. b.</em> system were analyzed. Canopy spectra were analyzed to estimate <em>Cr</em> and <em>Ni</em> contents using optimized spectral indices and LSSVM-based full-band modeling. Finally, <em>Cr</em> and <em>Ni</em> contents in <em>S. b.</em> leaves were estimated using the optimal inversion model, and further inversions based on enrichment characteristics were performed. The results were as follows. (1) In the May samples, the average soil <em>Cr</em> and <em>Ni</em> contents were 59.35 and 19.87 mg/kg, respectively, which were lower than the soil background values in Beijing and Hebei. <em>Cr</em> and <em>Ni</em> contents in roots, stems and leaves (RSL) were 26.96, 34.81, 37.09 mg/kg and 2.66, 4.08, 3.89 mg/kg, respectively. From June to September, the average <em>Cr</em> and <em>Ni</em> contents in the RSL of <em>S. b.</em> were 1.40, 0.76, 1.09 mg/kg and 1.11, 0.70, 1.23 mg/kg, respectively. (2) A strong correlation existed between low concentrations of HM in the soil and <em>S. b.</em> different parts, reaching a highly significant level (p < 0.01), supporting the use of <em>S. b.</em> leaves to estimate <em>Cr</em> and <em>Ni</em> in different soil parts and <em>S. b..</em> (3) For <em>Cr</em> and <em>Ni</em>, by modeling the preprocessed first derivative (FD) and second derivative (SD) spectra, the R<em>c</em> <sup>2</sup> = 0.98, RMSE<em>c</em> = 0.10 mg/kg, R<em>v</em> <sup>2</sup> = 0.55, and RMSE<em>v</em> = 0.33 mg/kg for <em>Cr</em>, and R<em>c</em> <sup>2</sup> = 0.99, RMSE<em>c</em> = 0.04 mg/kg, R<em>v</em> <sup>2</sup> = 0.48, and RMSE<em>v</em> = 0.34 mg/kg for <em>Ni</em> were obtained, demonstrating their strong ability to estimate <em>Cr</em> and <em>Ni</em> in <em>S. b.</em> leaves. (4) Using hyperspectral estimation of <em>Cr</em> and <em>Ni</em> in <em>S. b.</em> leaves, together with grouped inversion of enrichment characteristics, stems and roots exceeded 0.33. Therefore, canopy spectral analysis combined with enrichment patterns offers a practical method for monitoring <em>Cr</em> and <em>Ni</em> in Huangqin system, supporting the safety testing of Chinese herbal medicines.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105084"},"PeriodicalIF":8.6,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977870","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 : 2026-01-16DOI: 10.1016/j.jag.2026.105102
Zhihao Jiang, Liang Li, Xiuyun Shi, Weitianhua Ma, He Wang
Snow depth estimation using Global Navigation Satellite System-Reflectometry (GNSS-R) has emerged as a critical tool for monitoring global water environment dynamics. However, traditional linear GNSS-R snow depth estimation methods are often constrained by residual nonlinear errors induced by environmental factors and complex multipath effects. While recent studies have explored machine learning (ML) techniques like Support Vector Machines (SVM) and Random Forests (RF), their direct snow depth retrieval approaches are still susceptible to residual nonlinear errors during snow-free periods. To address these limitations, we proposes the Dual-Branch ML framework with decoupled architectures for nonlinear interference mitigation in GNSS-R-derived snow depth estimation. The Multi-Layer Perceptron (MLP), SVM, RF are employed for efficient snow state detection, and the 1D Convolutional Neural Network (CNN), Support Vector Regression (SVR), RF then leverage the extracted features (frequency, amplitude, phase, and previous day’s snow depth) to perform the precise regression task for snow depth estimation, respectively. Experimental results demonstrate significant improvements: the proposed method achieves an average root mean square error (RMSE) of 5.41 cm for the P350 station and 1.89 cm for the AB33 station, with correlation coefficients of 0.995 and 0.999, respectively. This approach not only effectively accounts for nonlinearities in GNSS-R snow depth estimation but also significantly enhances estimation accuracy, offering a robust and promising solution for global snow depth retrieval.
{"title":"Dual-Branch Machine Learning framework with decoupled architectures for nonlinear interference mitigation in GNSS-R snow depth estimation","authors":"Zhihao Jiang, Liang Li, Xiuyun Shi, Weitianhua Ma, He Wang","doi":"10.1016/j.jag.2026.105102","DOIUrl":"10.1016/j.jag.2026.105102","url":null,"abstract":"<div><div>Snow depth estimation using Global Navigation Satellite System-Reflectometry (GNSS-R) has emerged as a critical tool for monitoring global water environment dynamics. However, traditional linear GNSS-R snow depth estimation methods are often constrained by residual nonlinear errors induced by environmental factors and complex multipath effects. While recent studies have explored machine learning (ML) techniques like Support Vector Machines (SVM) and Random Forests (RF), their direct snow depth retrieval approaches are still susceptible to residual nonlinear errors during snow-free periods. To address these limitations, we proposes the Dual-Branch ML framework with decoupled architectures for nonlinear interference mitigation in GNSS-R-derived snow depth estimation. The Multi-Layer Perceptron (MLP), SVM, RF are employed for efficient snow state detection, and the 1D Convolutional Neural Network (CNN), Support Vector Regression (SVR), RF then leverage the extracted features (frequency, amplitude, phase, and previous day’s snow depth) to perform the precise regression task for snow depth estimation, respectively. Experimental results demonstrate significant improvements: the proposed method achieves an average root mean square error (RMSE) of 5.41 cm for the P350 station and 1.89 cm for the AB33 station, with correlation coefficients of 0.995 and 0.999, respectively. This approach not only effectively accounts for nonlinearities in GNSS-R snow depth estimation but also significantly enhances estimation accuracy, offering a robust and promising solution for global snow depth retrieval.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105102"},"PeriodicalIF":8.6,"publicationDate":"2026-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977787","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 : 2026-01-15DOI: 10.1016/j.jag.2026.105095
Cristopher Castro Traba , David Rijlaarsdam , Jian Guo , Roberto Del Prete , Gabriele Meoni
The rapid spread and destructive nature of wildfires and volcanic activity have intensified the need for low latency detection systems. The growing intensity and frequency of globally distributed thermal hotspots have driven the development of satellite-based detection solutions. Conventional approaches rely on ground-based processing, which limits low latency capabilities due to revisit times over ground stations and data handling requirements. This work proposes the first onboard payload processing pipeline for segmentation of thermal hotspots in raw multispectral satellite imagery. The pipeline leverages the Near InfraRed (NIR) and Short-Wave InfraRed (SWIR) spectral bands, and the combination of onboard Artificial Intelligence (AI) and raw imagery significantly reduces the delay between image acquisition and event detection. Furthermore, we present Segmentation of Thermal Hotspots in Raw Sentinel-2 data (SegTHRawS), the first publicly available dataset for thermal hotspot segmentation in raw multispectral satellite imagery. The segmentation model employed is a Fully Convolutional Network (FCN) derived from U-Net, named ResUnet-S2, designed for fast on-device inference. This model achieved an Intersection over Union (IoU) of 0.988 and an F-1 score of 0.986 on SegTHRawS, with its detection and generalization capabilities validated using an external thermal hotspot segmentation dataset. The proposed pipeline was verified on CubeSat-compatible hardware, achieving an end-to-end execution, from image acquisition to event detection, in 1.45 s, faster than the image acquisition process, and consuming a peak power of 4.05 W. These results demonstrate the potential of onboard processing solutions for minimizing the detection latency of current approaches, particularly for thermal hotspot segmentation, using edge computing satellite hardware.
{"title":"Towards onboard thermal hotspots segmentation with raw multispectral satellite imagery","authors":"Cristopher Castro Traba , David Rijlaarsdam , Jian Guo , Roberto Del Prete , Gabriele Meoni","doi":"10.1016/j.jag.2026.105095","DOIUrl":"10.1016/j.jag.2026.105095","url":null,"abstract":"<div><div>The rapid spread and destructive nature of wildfires and volcanic activity have intensified the need for low latency detection systems. The growing intensity and frequency of globally distributed thermal hotspots have driven the development of satellite-based detection solutions. Conventional approaches rely on ground-based processing, which limits low latency capabilities due to revisit times over ground stations and data handling requirements. This work proposes the first onboard payload processing pipeline for segmentation of thermal hotspots in raw multispectral satellite imagery. The pipeline leverages the Near InfraRed (NIR) and Short-Wave InfraRed (SWIR) spectral bands, and the combination of onboard Artificial Intelligence (AI) and raw imagery significantly reduces the delay between image acquisition and event detection. Furthermore, we present Segmentation of Thermal Hotspots in Raw Sentinel-2 data (SegTHRawS), the first publicly available dataset for thermal hotspot segmentation in raw multispectral satellite imagery. The segmentation model employed is a Fully Convolutional Network (FCN) derived from U-Net, named ResUnet-S2, designed for fast on-device inference. This model achieved an Intersection over Union (IoU) of 0.988 and an F-1 score of 0.986 on SegTHRawS, with its detection and generalization capabilities validated using an external thermal hotspot segmentation dataset. The proposed pipeline was verified on CubeSat-compatible hardware, achieving an end-to-end execution, from image acquisition to event detection, in 1.45 s, faster than the image acquisition process, and consuming a peak power of 4.05 W. These results demonstrate the potential of onboard processing solutions for minimizing the detection latency of current approaches, particularly for thermal hotspot segmentation, using edge computing satellite hardware.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105095"},"PeriodicalIF":8.6,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977789","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 : 2026-01-14DOI: 10.1016/j.jag.2026.105105
Qianhuizi Guo , Liangzhi Li , Ling Han
Accurate mapping of forest (F) and non-forest (NF) areas is essential for ecological assessment, resource management, and deforestation monitoring. However, complex backgrounds, severe class imbalance and redundant features continue to limit the accuracy and efficiency of network segmentation. To overcome these issues, we present ForResANeXt, a novel semantic segmentation network that uses Sentinel-2 multispectral imagery for forest/non-forest mapping. The model incorporates an AResCAB to enrich contextual feature representations while reducing redundancy and a lightweight embedded attention module to improve positional awareness. Furthermore, attention-gated skip connections suppress background noise and emphasize key spatial information, and a Focal Dice Loss function mitigates the impact of severe class imbalance. Experimental results demonstrate that ForResANeXt achieves a mIoU of 95.31%, surpassing U-Net and mainstream CNN variants in recall and F1 score for the minority non-forest class. It also outperforms several representative advanced CNN architectures and Transformer-based models in terms of Boundary IoU and Small Object Recall. Qualitative comparisons further confirm its superior capability in preserving structural details and delineating complex boundaries with reduced misclassification. Cross-regional transfer experiments validate the model’s robustness and generalization capability across diverse geographical and temporal conditions, and ablation studies confirm the effectiveness of each proposed component. Overall, ForResANeXt shows great promise for efficient and accurate forest cover mapping using multispectral satellite data.
{"title":"ForResANeXt: Forest/non-forest segmentation with aggregated residual attention network in satellite imagery","authors":"Qianhuizi Guo , Liangzhi Li , Ling Han","doi":"10.1016/j.jag.2026.105105","DOIUrl":"10.1016/j.jag.2026.105105","url":null,"abstract":"<div><div>Accurate mapping of forest (F) and non-forest (NF) areas is essential for ecological assessment, resource management, and deforestation monitoring. However, complex backgrounds, severe class imbalance and redundant features continue to limit the accuracy and efficiency of network segmentation. To overcome these issues, we present ForResANeXt, a novel semantic segmentation network that uses Sentinel-2 multispectral imagery for forest/non-forest mapping. The model incorporates an AResCAB to enrich contextual feature representations while reducing redundancy and a lightweight embedded attention module to improve positional awareness. Furthermore, attention-gated skip connections suppress background noise and emphasize key spatial information, and a Focal Dice Loss function mitigates the impact of severe class imbalance. Experimental results demonstrate that ForResANeXt achieves a mIoU of 95.31%, surpassing U-Net and mainstream CNN variants in recall and F1 score for the minority non-forest class. It also outperforms several representative advanced CNN architectures and Transformer-based models in terms of Boundary IoU and Small Object Recall. Qualitative comparisons further confirm its superior capability in preserving structural details and delineating complex boundaries with reduced misclassification. Cross-regional transfer experiments validate the model’s robustness and generalization capability across diverse geographical and temporal conditions, and ablation studies confirm the effectiveness of each proposed component. Overall, ForResANeXt shows great promise for efficient and accurate forest cover mapping using multispectral satellite data.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105105"},"PeriodicalIF":8.6,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977788","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 : 2026-01-13DOI: 10.1016/j.jag.2026.105092
Jie Li , Ziyi Zhang , Tongwen Li , Qiangqiang Yuan , Liangpei Zhang
Carbon dioxide (CO2) is a dominant greenhouse gas and has a considerable effect on climate change. Satellite remote sensing is commonly used to acquire atmospheric CO2 concentrations. However, the limited spatial coverage of a single satellite makes the obtainment of full-coverage CO2 data difficult. In this study, a daily dataset of global seamless column-averaged dry-air mole fractions of CO2 (XCO2) was generated with a high spatial resolution of 0.1° from 2016 to 2020, by using a stacking machine learning method. The proposed XCO2 dataset shows a satisfactory performance, with a root mean square error (RMSE) of 0.9697 ppm and correlation coefficient (R) of 0.9868 in the 10-fold cross validation. The spatial validation reveals good generalization ability, with continent-by-continent validation results showing an R greater than 0.93. The proposed dataset reports high consistency and accuracy in the ground-based validation, with an RMSE of 1.0855 ppm. Out of 24 stations, 22 demonstrate a precision of R greater than 0.95. In comparison with two XCO2 model simulations, our reconstructions show a better consistency with ground observations. Spatial analyses at continent, national, and Chinese provincial levels, and temporal trends at daily, monthly, seasonal, and annual scales, are provided. Furthermore, benefitting from the daily temporal resolution, two typical examples of wildfire events, namely the Fort McMurray wildfire and the Blue Cut Fire, are evaluated. Our dataset can effectively capture fine-scale XCO2 variations and has the potential to characterize carbon sources and sinks. The dataset can be obtained freely at https://zenodo.org/records/15191247.
{"title":"Global daily seamless XCO2 Mapping (2016–2020): Spatio-temporal trends and variations during wildfire events","authors":"Jie Li , Ziyi Zhang , Tongwen Li , Qiangqiang Yuan , Liangpei Zhang","doi":"10.1016/j.jag.2026.105092","DOIUrl":"10.1016/j.jag.2026.105092","url":null,"abstract":"<div><div>Carbon dioxide (CO<sub>2</sub>) is a dominant greenhouse gas and has a considerable effect on climate change. Satellite remote sensing is commonly used to acquire atmospheric CO<sub>2</sub> concentrations. However, the limited spatial coverage of a single satellite makes the obtainment of full-coverage CO<sub>2</sub> data difficult. In this study, a daily dataset of global seamless column-averaged dry-air mole fractions of CO<sub>2</sub> (XCO<sub>2</sub>) was generated with a high spatial resolution of 0.1° from 2016 to 2020, by using a stacking machine learning method. The proposed XCO<sub>2</sub> dataset shows a satisfactory performance, with a root mean square error (RMSE) of 0.9697 ppm and correlation coefficient (R) of 0.9868 in the 10-fold cross validation. The spatial validation reveals good generalization ability, with continent-by-continent validation results showing an R greater than 0.93. The proposed dataset reports high consistency and accuracy in the ground-based validation, with an RMSE of 1.0855 ppm. Out of 24 stations, 22 demonstrate a precision of R greater than 0.95. In comparison with two XCO<sub>2</sub> model simulations, our reconstructions show a better consistency with ground observations. Spatial analyses at continent, national, and Chinese provincial levels, and temporal trends at daily, monthly, seasonal, and annual scales, are provided. Furthermore, benefitting from the daily temporal resolution, two typical examples of wildfire events, namely the Fort McMurray wildfire and the Blue Cut Fire, are evaluated. Our dataset can effectively capture fine-scale XCO<sub>2</sub> variations and has the potential to characterize carbon sources and sinks. The dataset can be obtained freely at <span><span>https://zenodo.org/records/15191247</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105092"},"PeriodicalIF":8.6,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957311","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}
Urban vegetation, providing critical ecosystem services and supporting biodiversity, is essential for sustainable and resilient cities. Yet its semantic representation in urban digital twins and 3D city models remains inadequate for advanced modeling. Open data standards facilitate interoperable urban modeling. However, the CityGML standard, despite its widespread adoption, provides limited semantic depth for vegetation in its current Vegetation module, constraining dynamic and interdisciplinary applications.
To address this, we present the international stakeholder survey evaluation alongside the updated version of the conceptual model of the CityGML 3.0 Vegetation Application Domain Extension (Vegetation ADE v1.1). The ADE enhances semantic richness by introducing extended attributes, feature types, enumerations, and code lists for both SolitaryVegetationObject and PlantCover, including structural components (crown, trunk, root) and dynamic properties via the CityGML Dynamizer module. Additionally, a vegetation management class is established.
Refined through feedback from an international stakeholder survey, the model reflects practical requirements from domains including urban studies, ecology, and public administration. The core data modeling structure of ADE v1.0, unanimously supported by survey participants, was retained for ADE v1.1. The new version introduces substantial improvements: 15 attributes were modified, 12 added, and one removed. Nine code lists were revised and one added. Two enumerations were updated. These enhancements ensure ADE v1.1 achieves improved semantic clarity and usability, supporting diverse applications such as vegetation monitoring and adaptive green infrastructure management. This work demonstrates a replicable methodology for participatory standard development and advances the integration of vegetation into data-driven urban digital twins and 3D city models.
{"title":"Urban vegetation semantics in CityGML: Key stakeholder survey findings and vegetation ADE development","authors":"Laura Mrosla , Dessislava Petrova-Antonova , Simeon Malinov , Henna Fabritius","doi":"10.1016/j.jag.2025.105043","DOIUrl":"10.1016/j.jag.2025.105043","url":null,"abstract":"<div><div>Urban vegetation, providing critical ecosystem services and supporting biodiversity, is essential for sustainable and resilient cities. Yet its semantic representation in urban digital twins and 3D city models remains inadequate for advanced modeling. Open data standards facilitate interoperable urban modeling. However, the CityGML standard, despite its widespread adoption, provides limited semantic depth for vegetation in its current Vegetation module, constraining dynamic and interdisciplinary applications.</div><div>To address this, we present the international stakeholder survey evaluation alongside the updated version of the conceptual model of the CityGML 3.0 Vegetation Application Domain Extension (Vegetation ADE v1.1). The ADE enhances semantic richness by introducing extended attributes, feature types, enumerations, and code lists for both <em>SolitaryVegetationObject</em> and <em>PlantCover</em>, including structural components (crown, trunk, root) and dynamic properties via the CityGML <em>Dynamizer</em> module. Additionally, a vegetation management class is established.</div><div>Refined through feedback from an international stakeholder survey, the model reflects practical requirements from domains including urban studies, ecology, and public administration. The core data modeling structure of ADE v1.0, unanimously supported by survey participants, was retained for ADE v1.1. The new version introduces substantial improvements: 15 attributes were modified, 12 added, and one removed. Nine code lists were revised and one added. Two enumerations were updated. These enhancements ensure ADE v1.1 achieves improved semantic clarity and usability, supporting diverse applications such as vegetation monitoring and adaptive green infrastructure management. This work demonstrates a replicable methodology for participatory standard development and advances the integration of vegetation into data-driven urban digital twins and 3D city models.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105043"},"PeriodicalIF":8.6,"publicationDate":"2026-01-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957312","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