International journal of applied earth observation and geoinformation : ITC journal最新文献_第3页

Mapping Spatio-Temporal dynamics of irrigated agriculture in Nepal using MODIS NDVI and statistical data with Google Earth Engine: A step towards improved irrigation planning 利用MODIS NDVI和谷歌Earth Engine的统计数据绘制尼泊尔灌溉农业的时空动态：朝着改善灌溉规划迈出的一步

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2024.104345

Pramit Ghimire , Saroj Karki , Vishnu Prasad Pandey , Ananta Man Singh Pradhan

The importance of water resources in supporting food production is ever increasing, especially in the face of climate change, urbanization and population growth. This study aims to map and analyze the spatio-temporal dynamics of irrigated agricultural areas to support improved planning of irrigation water and irrigation sector in Nepal. Using the Normalized Difference Vegetation Index (NDVI) from the Moderate Resolution Imaging Spectroradiometer (MODIS) employing Google Earth Engine (GEE) platform, this study classifies and analyzes change in irrigated and rainfed areas over the past two decades. NDVI time series analysis across different physiographic regions uncovered two cropping cycles annually in the Terai and Siwalik regions. In contrast, predominantly a single cropping cycle was observed in the Middle and High Mountain regions. The k-means clustering algorithm was applied to NDVI time series within the agriculture land use database of the International Centre for Integrated Mountain Development (ICIMOD) for Nepal. The obtained irrigated areas distribution were also analyzed across different provinces of Nepal as provinces are the main functional administrative divisions after federal level that are responsible for irrigation development. The produced irrigation areas distribution showed reasonable accuracy as compared to the statistical irrigation areas database of the Department of Water Resources and Irrigation (DWRI), Nepal. The results showed that, on average, approximately 60% (2.18 million hectares) of agricultural land was irrigated annually over the past decade. The findings will provide valuable insights for sustainable irrigation and water resource management, crop productivity enhancement, and strategy formulation to ensure food and water security in Nepal.

水资源在支持粮食生产方面的重要性日益增加，特别是在面对气候变化、城市化和人口增长的情况下。本研究旨在绘制和分析灌溉农业区的时空动态，以支持尼泊尔灌溉用水和灌溉部门的改进规划。利用谷歌Earth Engine （GEE）平台的MODIS中分辨率植被指数（NDVI），对近20年来中国灌区和雨牧区的变化进行了分类分析。不同地理区域的NDVI时间序列分析发现，Terai和Siwalik地区每年有两个种植周期。而在中、高山地区，主要是单作周期。将k-means聚类算法应用于国际山地综合发展中心（ICIMOD）尼泊尔农业用地数据库内的NDVI时间序列。获得的灌溉区分布也在尼泊尔不同省份之间进行了分析，因为省份是联邦一级之后负责灌溉发展的主要职能行政区划。与尼泊尔水资源和灌溉部（DWRI）的统计灌溉区数据库相比，生产灌溉区分布显示出合理的准确性。结果表明，在过去十年中，平均每年约有60%（218万公顷）的农业用地得到灌溉。研究结果将为尼泊尔的可持续灌溉和水资源管理、提高作物生产力和制定战略提供有价值的见解，以确保粮食和水安全。

{"title":"Mapping Spatio-Temporal dynamics of irrigated agriculture in Nepal using MODIS NDVI and statistical data with Google Earth Engine: A step towards improved irrigation planning","authors":"Pramit Ghimire , Saroj Karki , Vishnu Prasad Pandey , Ananta Man Singh Pradhan","doi":"10.1016/j.jag.2024.104345","DOIUrl":"10.1016/j.jag.2024.104345","url":null,"abstract":"<div><div>The importance of water resources in supporting food production is ever increasing, especially in the face of climate change, urbanization and population growth. This study aims to map and analyze the spatio-temporal dynamics of irrigated agricultural areas to support improved planning of irrigation water and irrigation sector in Nepal. Using the Normalized Difference Vegetation Index (NDVI) from the Moderate Resolution Imaging Spectroradiometer (MODIS) employing Google Earth Engine (GEE) platform, this study classifies and analyzes change in irrigated and rainfed areas over the past two decades. NDVI time series analysis across different physiographic regions uncovered two cropping cycles annually in the Terai and Siwalik regions. In contrast, predominantly a single cropping cycle was observed in the Middle and High Mountain regions. The k-means clustering algorithm was applied to NDVI time series within the agriculture land use database of the International Centre for Integrated Mountain Development (ICIMOD) for Nepal. The obtained irrigated areas distribution were also analyzed across different provinces of Nepal as provinces are the main functional administrative divisions after federal level that are responsible for irrigation development. The produced irrigation areas distribution showed reasonable accuracy as compared to the statistical irrigation areas database of the Department of Water Resources and Irrigation (DWRI), Nepal. The results showed that, on average, approximately 60% (2.18 million hectares) of agricultural land was irrigated annually over the past decade. The findings will provide valuable insights for sustainable irrigation and water resource management, crop productivity enhancement, and strategy formulation to ensure food and water security in Nepal.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104345"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142929500","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CUG-STCN: A seabed topography classification framework based on knowledge graph-guided vision mamba network

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104383

Haoyi Wang , Weitao Chen , Xianju Li , Qianyong Liang , Xuwen Qin , Jun Li

Multibeam sounding is a high-precision remote sensing method for seabed detection. Seabed topography classification is crucial for marine science research, resource exploration and engineering. When using multibeam data for seabed topography automatic classification, the fuzzy boundaries of different topographic entities, redundancy of multimodal data, and the lack of geological knowledge guidance have led to low classification accuracy. Thus, a knowledge graph-guided vision mamba seabed topography classification network (CUG-STCN) was constructed, consisting of three modules: (1) The long sequence modeling mamba-based encoder addresses the fuzzy seabed topography boundary. It uses 2D-selective-scan to create image blocks in different scanning directions. By combining with the selective state space model to capture long-range dependencies and ensure transmission of spatial context information while maintaining linear computational complexity. (2) The cross-modal information interaction and fusion module addresses the redundancy of multimodal information. By employing a bidirectional information interaction mechanism, it captures the correlations of seabed topography between different modalities and achieving feature fusion. (3) The seabed topography knowledge graph-guided semantic perception module guides the geological knowledge. It constructs seabed topography knowledge vectors through entity query and word embedding, using the similarity between vectors to create a similarity measurement matrix. It provides geological knowledge, enhancing the modeling capability of complex seabed topography relationship. CUG-STCN achieves OA of 90.11% and mIOU of 48.50%, outperforming six mainstream networks, which at most, achieve the OA and mIOU improvements of 5.37% and 14.18%. Notably, the application of CUG-STCN in other regions demonstrates its strong generalization performance.

{"title":"CUG-STCN: A seabed topography classification framework based on knowledge graph-guided vision mamba network","authors":"Haoyi Wang , Weitao Chen , Xianju Li , Qianyong Liang , Xuwen Qin , Jun Li","doi":"10.1016/j.jag.2025.104383","DOIUrl":"10.1016/j.jag.2025.104383","url":null,"abstract":"<div><div>Multibeam sounding is a high-precision remote sensing method for seabed detection. Seabed topography classification is crucial for marine science research, resource exploration and engineering. When using multibeam data for seabed topography automatic classification, the fuzzy boundaries of different topographic entities, redundancy of multimodal data, and the lack of geological knowledge guidance have led to low classification accuracy. Thus, a knowledge graph-guided vision mamba seabed topography classification network (CUG-STCN) was constructed, consisting of three modules: (1) The long sequence modeling mamba-based encoder addresses the fuzzy seabed topography boundary. It uses 2D-selective-scan to create image blocks in different scanning directions. By combining with the selective state space model to capture long-range dependencies and ensure transmission of spatial context information while maintaining linear computational complexity. (2) The cross-modal information interaction and fusion module addresses the redundancy of multimodal information. By employing a bidirectional information interaction mechanism, it captures the correlations of seabed topography between different modalities and achieving feature fusion. (3) The seabed topography knowledge graph-guided semantic perception module guides the geological knowledge. It constructs seabed topography knowledge vectors through entity query and word embedding, using the similarity between vectors to create a similarity measurement matrix. It provides geological knowledge, enhancing the modeling capability of complex seabed topography relationship. CUG-STCN achieves OA of 90.11% and mIOU of 48.50%, outperforming six mainstream networks, which at most, achieve the OA and mIOU improvements of 5.37% and 14.18%. Notably, the application of CUG-STCN in other regions demonstrates its strong generalization performance.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104383"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143050023","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Detecting tropical freshly-opened swidden fields using a combined algorithm of continuous change detection and support vector machine

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104403

Ningsang Jiang , Peng Li , Zhiming Feng

Swidden agriculture, widely practiced by impoverished ethnic groups, continues to undergo rapid transition and transformation in tropical highlands. Exploring universal approaches for accurate mapping of newly-opened swiddens and fallows of different ages has not yet been stopped. The development of data-, information-, and knowledge-based algorithms for monitoring swidden agriculture requires integration of multi-dimensional features. The first part of the Continuous Change Detection and Classification (CCDC) algorithm holds promising potential in capturing abrupt changes. However, the CCD-derived temporal attributes and other multi-dimension features are seldom utilized to monitor swidden agriculture. Here, a combined algorithm integrating CCD and Support Vector Machine (SVM) is firstly developed to comprehensively highlight fundamental characteristics of swidden agriculture for maximumly and effectively mapping freshly opened swiddens. Local experimental results demonstrate that the CCD-SVM algorithm significantly enhances the performance of SVM in newly-opened swidden identification, with an average accuracy of over 85% (around a 10–20% improvement) under different land cover conditions. Next, CCD-SVM is applied to generate the 2019 map of newly-opened swidden in Laos using Landsat-8 dry-season (February to April) imagery. Comparisons with the same year results obtained from the CCDC-Spectral Mixture Analysis (SMA) show that CCD-SVM (94.69%) outperforms CCDC-SMA (87.52%) primarily due to less commission errors. Features inclusion of terrain and fire greatly improves classification accuracy. Additionally, over 60% of Laotian swiddens cross-validated by the 375-meter Visible Infrared Imaging Radiometer Suite active fires demonstrate CCD-SVM’s reliability and fidelity. The integration CCDC with SVM represents a novelty in combining time series analysis and machine learning techniques and helps monitor annual swidden agriculture in the tropics.

{"title":"Detecting tropical freshly-opened swidden fields using a combined algorithm of continuous change detection and support vector machine","authors":"Ningsang Jiang , Peng Li , Zhiming Feng","doi":"10.1016/j.jag.2025.104403","DOIUrl":"10.1016/j.jag.2025.104403","url":null,"abstract":"<div><div>Swidden agriculture, widely practiced by impoverished ethnic groups, continues to undergo rapid transition and transformation in tropical highlands. Exploring universal approaches for accurate mapping of newly-opened swiddens and fallows of different ages has not yet been stopped. The development of data-, information-, and knowledge-based algorithms for monitoring swidden agriculture requires integration of multi-dimensional features. The first part of the Continuous Change Detection and Classification (CCDC) algorithm holds promising potential in capturing abrupt changes. However, the CCD-derived temporal attributes and other multi-dimension features are seldom utilized to monitor swidden agriculture. Here, a combined algorithm integrating CCD and Support Vector Machine (SVM) is firstly developed to comprehensively highlight fundamental characteristics of swidden agriculture for maximumly and effectively mapping freshly opened swiddens. Local experimental results demonstrate that the CCD-SVM algorithm significantly enhances the performance of SVM in newly-opened swidden identification, with an average accuracy of over 85% (around a 10–20% improvement) under different land cover conditions. Next, CCD-SVM is applied to generate the 2019 map of newly-opened swidden in Laos using Landsat-8 dry-season (February to April) imagery. Comparisons with the same year results obtained from the CCDC-Spectral Mixture Analysis (SMA) show that CCD-SVM (94.69%) outperforms CCDC-SMA (87.52%) primarily due to less commission errors. Features inclusion of terrain and fire greatly improves classification accuracy. Additionally, over 60% of Laotian swiddens cross-validated by the 375-meter Visible Infrared Imaging Radiometer Suite active fires demonstrate CCD-SVM’s reliability and fidelity. The integration CCDC with SVM represents a novelty in combining time series analysis and machine learning techniques and helps monitor annual swidden agriculture in the tropics.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104403"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143211675","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SSMM: Semi-supervised manifold method with spatial-spectral self-training and regularized metric constraints for hyperspectral image dimensionality reduction

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104373

Bei Zhu , Yao Jin , Xuehua Guan , Yanni Dong

Manifold learning is an important technique for dimensionality reduction in hyperspectral images. It maps data from high dimensions to low dimensions to eliminate redundant information. However, the existing manifold learning methods cannot effectively solve the problem of lacking label information and ignore the negative impact of dimensionality reduction on sample division. To address these, we propose a semi-supervised manifold method with spatial-spectral self-training and regularized metric constraints (SSMM) for hyperspectral image dimensionality reduction. The spatial-spectral self-training module is proposed, which learns pseudo-labels by jointly training the spatial and spectral information. This module first locates the spatial neighbors of the labeledit can adapt to different data distributions and feature samples and then sets an adaptive threshold based on the spectral features of labeled samples to filter spatial neighbors, so as to obtain the spatial-spectral neighbors as pseudo-labeled samples. In addition, to divide the sample categories while dimensionality reduction, low-dimensional manifold embedding is constructed and the metric constraint is imposed on the manifolds. Specifically, the Gaussian kernel function based on Mahalanobis distance is used to map the data into a more discriminative low-dimensional manifold embedding. At the same time, the regularized distance metric constraint is imposed on the manifold, so that samples of the same class are clustered and different classes are mutually exclusive. SSMM conducts various forms of experiments on the Houston 2013, Indian Pines, and Washington DC datasets. In the dimensionality reduction experiments, the overall accuracy of SSMM in any dimension is higher than that of other algorithms. In the classification experiments, the KAPPA coefficient of SSMM on the three data sets is improved by 1.41%, 0.61%, and 0.27% respectively. The feature extraction experiments show superior clustering performance. These experimental results demonstrate that SSMM not only effectively solves the problem of insufficient label information, but also significantly improves the classification accuracy of hyperspectral images after dimensionality reduction, which is superior to the existing manifold learning methods.

{"title":"SSMM: Semi-supervised manifold method with spatial-spectral self-training and regularized metric constraints for hyperspectral image dimensionality reduction","authors":"Bei Zhu , Yao Jin , Xuehua Guan , Yanni Dong","doi":"10.1016/j.jag.2025.104373","DOIUrl":"10.1016/j.jag.2025.104373","url":null,"abstract":"<div><div>Manifold learning is an important technique for dimensionality reduction in hyperspectral images. It maps data from high dimensions to low dimensions to eliminate redundant information. However, the existing manifold learning methods cannot effectively solve the problem of lacking label information and ignore the negative impact of dimensionality reduction on sample division. To address these, we propose a semi-supervised manifold method with spatial-spectral self-training and regularized metric constraints (SSMM) for hyperspectral image dimensionality reduction. The spatial-spectral self-training module is proposed, which learns pseudo-labels by jointly training the spatial and spectral information. This module first locates the spatial neighbors of the labeledit can adapt to different data distributions and feature samples and then sets an adaptive threshold based on the spectral features of labeled samples to filter spatial neighbors, so as to obtain the spatial-spectral neighbors as pseudo-labeled samples. In addition, to divide the sample categories while dimensionality reduction, low-dimensional manifold embedding is constructed and the metric constraint is imposed on the manifolds. Specifically, the Gaussian kernel function based on Mahalanobis distance is used to map the data into a more discriminative low-dimensional manifold embedding. At the same time, the regularized distance metric constraint is imposed on the manifold, so that samples of the same class are clustered and different classes are mutually exclusive. SSMM conducts various forms of experiments on the Houston 2013, Indian Pines, and Washington DC datasets. In the dimensionality reduction experiments, the overall accuracy of SSMM in any dimension is higher than that of other algorithms. In the classification experiments, the KAPPA coefficient of SSMM on the three data sets is improved by 1.41%, 0.61%, and 0.27% respectively. The feature extraction experiments show superior clustering performance. These experimental results demonstrate that SSMM not only effectively solves the problem of insufficient label information, but also significantly improves the classification accuracy of hyperspectral images after dimensionality reduction, which is superior to the existing manifold learning methods.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104373"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143350537","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Dual Fine-Grained network with frequency Transformer for change detection on remote sensing images

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104393

Zhen Li , Zhenxin Zhang , Mengmeng Li , Liqiang Zhang , Xueli Peng , Rixing He , Leidong Shi

Change detection is a fundamental yet challenging task in remote sensing, crucial for monitoring urban expansion, land use changes, and environmental dynamics. However, compared with common color images, objects in remote sensing images exhibit minimal interclass variation and significant intraclass variation in the spectral dimension, with obvious scale inconsistency in the spatial dimension. Change detection complexity presents significant challenges, including differentiating similar objects, accounting for scale variations, and identifying pseudo changes. This research introduces a dual fine-grained network with a frequency Transformer (named as FTransDF-Net) to address the above issues. Specifically, for small-scale and approximate spectral ground objects, the network employs an encoder-decoder architecture consisting of dual fine-grained gated (DFG) modules. This enables the extraction and fusion of fine-grained level information in dual dimensions of features, facilitating a comprehensive analysis of their differences and correlations. As a result, a dynamic fusion representation of salient information is achieved. Additionally, we develop a lightweight frequency transformer (LFT) with minimal parameters for detecting large-scale ground objects that undergo significant changes over time. This is achieved by incorporating a frequency attention (FA) module, which utilizes Fourier transform to model long-range dependencies and combines global adaptive attentive features with multi-level fine-grained features. Our comparative experiments across four publicly available datasets demonstrate that FTransDF-Net reaches advanced results. Importantly, it outperforms the leading comparison method by 1.23% and 2.46% regarding IoU metrics concerning CDD and DSIFN, respectively. Furthermore, efficacy for each module is substantiated through ablation experiments. The code is accessible on https://github.com/LeeThrzz/FTrans-DF-Net.

{"title":"Dual Fine-Grained network with frequency Transformer for change detection on remote sensing images","authors":"Zhen Li , Zhenxin Zhang , Mengmeng Li , Liqiang Zhang , Xueli Peng , Rixing He , Leidong Shi","doi":"10.1016/j.jag.2025.104393","DOIUrl":"10.1016/j.jag.2025.104393","url":null,"abstract":"<div><div>Change detection is a fundamental yet challenging task in remote sensing, crucial for monitoring urban expansion, land use changes, and environmental dynamics. However, compared with common color images, objects in remote sensing images exhibit minimal interclass variation and significant intraclass variation in the spectral dimension, with obvious scale inconsistency in the spatial dimension. Change detection complexity presents significant challenges, including differentiating similar objects, accounting for scale variations, and identifying pseudo changes. This research introduces a dual fine-grained network with a frequency Transformer (named as FTransDF-Net) to address the above issues. Specifically, for small-scale and approximate spectral ground objects, the network employs an encoder-decoder architecture consisting of dual fine-grained gated (DFG) modules. This enables the extraction and fusion of fine-grained level information in dual dimensions of features, facilitating a comprehensive analysis of their differences and correlations. As a result, a dynamic fusion representation of salient information is achieved. Additionally, we develop a lightweight frequency transformer (LFT) with minimal parameters for detecting large-scale ground objects that undergo significant changes over time. This is achieved by incorporating a frequency attention (FA) module, which utilizes Fourier transform to model long-range dependencies and combines global adaptive attentive features with multi-level fine-grained features. Our comparative experiments across four publicly available datasets demonstrate that FTransDF-Net reaches advanced results. Importantly, it outperforms the leading comparison method by 1.23% and 2.46% regarding IoU metrics concerning CDD and DSIFN, respectively. Furthermore, efficacy for each module is substantiated through ablation experiments. The code is accessible on <span><span>https://github.com/LeeThrzz/FTrans-DF-Net</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104393"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143372456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

CSTN: A cross-region crop mapping method integrating self-training and contrastive domain adaptation

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104379

Shuwen Peng , Liqiang Zhang , Rongchang Xie , Ying Qu

Crop mapping is essential for agricultural management and food production monitoring, but challenges like limited crop labels and poor model generalization significantly hinder large-scale crop mapping. Here, we introduce a novel Contrastive Self-Training Network (CSTN), integrating a self-training strategy and contrastive domain adaptation (CDA) for cross-region crop mapping. CSTN uses pseudo-labels in the target region generated by the self-training strategy to assist supervised learning, and aligns features across regions using class-aware prototypes. Qualitative and quantitative evaluations demonstrate that CSTN significantly outperforms state-of-the-art methods with a 12.29 % increase in average F1-score, particularly in maize identification. Moreover, CSTN also enables early-season crop classification for pre-harvest decision-making applications. The interpretability of the model is demonstrated through an in-depth analysis of feature map visualizations, attention map visualizations, and the effectiveness of the modules. This study provides a robust method for enhancing large-scale crop mapping and facilitating more accurate and timely agricultural practices.

{"title":"CSTN: A cross-region crop mapping method integrating self-training and contrastive domain adaptation","authors":"Shuwen Peng , Liqiang Zhang , Rongchang Xie , Ying Qu","doi":"10.1016/j.jag.2025.104379","DOIUrl":"10.1016/j.jag.2025.104379","url":null,"abstract":"<div><div>Crop mapping is essential for agricultural management and food production monitoring, but challenges like limited crop labels and poor model generalization significantly hinder large-scale crop mapping. Here, we introduce a novel Contrastive Self-Training Network (CSTN), integrating a self-training strategy and contrastive domain adaptation (CDA) for cross-region crop mapping. CSTN uses pseudo-labels in the target region generated by the self-training strategy to assist supervised learning, and aligns features across regions using class-aware prototypes. Qualitative and quantitative evaluations demonstrate that CSTN significantly outperforms state-of-the-art methods with a 12.29 % increase in average F1-score, particularly in maize identification. Moreover, CSTN also enables early-season crop classification for pre-harvest decision-making applications. The interpretability of the model is demonstrated through an in-depth analysis of feature map visualizations, attention map visualizations, and the effectiveness of the modules. This study provides a robust method for enhancing large-scale crop mapping and facilitating more accurate and timely agricultural practices.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104379"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143377086","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SD-Mamba: A lightweight synthetic-decompression network for cross-modal flood change detection

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104409

Yu Shen, Shuang Yao, Zhenkai Qiang, Guanxiang Pei

Cross-modal flood change detection using optical and SAR images has become one of the most commonly used techniques for monitoring the progression of flooding events. Existing methods fail to adequately capture the interrelationship between semantics and changes, which limits the potential for effective flood detection. To address this issue, we propose a lightweight Synthetic-decompression network. The synthetic component is divided into four stages, each of which employs a Multi-branch Asymmetric Part-convolution block (MAPC) and a Temporal Semantic Interaction module (TSIM) to extract semantic features from dual-temporal images. Subsequently, these features are fed into the Temporal-mamba (T-Mamba), which uses 4D Selective Scanning (SS4D) to traverse temporal change information in four directions. The decompression component employs a three-stage Asymmetric Coordinate-convolution block (ACoord-Conv) to project the change results onto the source images, thereby indirectly supervising the model’s detection performance. Compared to the 22 state-of-the-art (SOTA) lightweight methods, SD-Mamba achieves an optimal balance between computational efficiency and detection accuracy. Under the same computational conditions, SD-Mamba demonstrated superior performance to other Mamba-based models, with an improvement of 1.01% in mIoU, while maintaining a lightweight structure with only 5.32M parameters and 12.24G floating-point operations (FLops). The code is available at https://github.com/yaoshuang-yaobo/SD-Mamba.

{"title":"SD-Mamba: A lightweight synthetic-decompression network for cross-modal flood change detection","authors":"Yu Shen, Shuang Yao, Zhenkai Qiang, Guanxiang Pei","doi":"10.1016/j.jag.2025.104409","DOIUrl":"10.1016/j.jag.2025.104409","url":null,"abstract":"<div><div>Cross-modal flood change detection using optical and SAR images has become one of the most commonly used techniques for monitoring the progression of flooding events. Existing methods fail to adequately capture the interrelationship between semantics and changes, which limits the potential for effective flood detection. To address this issue, we propose a lightweight Synthetic-decompression network. The synthetic component is divided into four stages, each of which employs a Multi-branch Asymmetric Part-convolution block (MAPC) and a Temporal Semantic Interaction module (TSIM) to extract semantic features from dual-temporal images. Subsequently, these features are fed into the Temporal-mamba (T-Mamba), which uses 4D Selective Scanning (SS4D) to traverse temporal change information in four directions. The decompression component employs a three-stage Asymmetric Coordinate-convolution block (ACoord-Conv) to project the change results onto the source images, thereby indirectly supervising the model’s detection performance. Compared to the 22 state-of-the-art (SOTA) lightweight methods, SD-Mamba achieves an optimal balance between computational efficiency and detection accuracy. Under the same computational conditions, SD-Mamba demonstrated superior performance to other Mamba-based models, with an improvement of 1.01% in mIoU, while maintaining a lightweight structure with only 5.32M parameters and 12.24G floating-point operations (FLops). The code is available at <span><span>https://github.com/yaoshuang-yaobo/SD-Mamba</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104409"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143402957","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Corrigendum to “L-band microwave-retrieved fuel temperature predicts million-hectare-scale destructive wildfires” [Int. J. Appl. Earth Obs. Geoinf. 129 (2024) 103776]

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2024.104299

Ju Hyoung Lee , Sander Veraverbeke , Brendan Rogers , Yann H. Kerr

引用次数: 0

Generating MODIS hourly land surface temperature under clear sky conditions using Fourier series analysis 利用傅立叶级数分析生成晴空条件下MODIS每小时地表温度

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2024.104341

Hadi Zare Khormizi , Mohammad Jafari , Hamidreza Ghafarian Malamiri , Ali Tavili , Hamidreza Keshtkar

Land surface temperature (LST) data with high temporal and spatial resolution are used in many studies, e.g. to assess climate changes, land–atmosphere interactions, surface energy balance, etc. However, clouds and the limitations of geostationary and polar orbiting satellites hinder the collection of high-quality thermal infrared (TIR) data. This research aims to generate hourly LST data from the Moderate Resolution Imaging Spectroradiometer (MODIS) with four daily observations. The Multi-channel Singular Spectrum Analysis (M−SSA) algorithm was used to reconstruct lost data due to clouds in the MODIS annual LST time series. Subsequently, Fourier series analysis was employed to generate hourly LST data based on the four MODIS observations per day. The developed Fourier series model was evaluated using hourly LST data from Meteosat-9 and ground surface soil temperature data at eight different Ameriflux sites. The evaluation of the Fourier series model showed that the Root Mean Square Error (RMSE) and coefficient of determination (R²) between the hourly LST data from the Meteosat-9 satellite and the hourly LST data generated by the Fourier series model using four simultaneous MODIS observations averaged 1.70 Kelvin and 0.98, respectively, throughout Iran. For Ameriflux sites, the average RMSE and R2 were 1.15 K and 0.98 between the surface soil temperature data and the surface soil temperature data generated using four simultaneous MODIS observations per day, respectively. Notably, the highest RMSE was observed during sunrise and sunset.

具有高时空分辨率的地表温度（LST）数据被用于许多研究，如评估气候变化、陆-气相互作用、地表能量平衡等。然而，云层以及地球静止轨道卫星和极轨卫星的局限性阻碍了高质量热红外数据的收集。本研究的目的是利用中分辨率成像光谱仪（MODIS）每日4次观测，生成每小时的地表温度数据。采用多通道奇异谱分析（M−SSA）算法重建MODIS年LST时间序列中由于云层造成的数据丢失。随后，基于MODIS每天4次观测数据，采用傅立叶级数分析生成逐时LST数据。利用Meteosat-9的每小时地表温度数据和八个不同Ameriflux站点的地表土壤温度数据，对开发的傅立叶级数模型进行了评估。傅里叶级数模型的评估表明，在伊朗，Meteosat-9卫星每小时地表温度数据与使用四个同时MODIS观测的傅里叶级数模型每小时地表温度数据之间的均方根误差（RMSE）和决定系数（R2）平均分别为1.70开尔文和0.98开尔文。Ameriflux站点的地表土壤温度数据与4次MODIS观测数据的平均RMSE和R2分别为1.15 K和0.98。值得注意的是，在日出和日落时观察到最高的RMSE。

{"title":"Generating MODIS hourly land surface temperature under clear sky conditions using Fourier series analysis","authors":"Hadi Zare Khormizi , Mohammad Jafari , Hamidreza Ghafarian Malamiri , Ali Tavili , Hamidreza Keshtkar","doi":"10.1016/j.jag.2024.104341","DOIUrl":"10.1016/j.jag.2024.104341","url":null,"abstract":"<div><div>Land surface temperature (LST) data with high temporal and spatial resolution are used in many studies, e.g. to assess climate changes, land–atmosphere interactions, surface energy balance, etc. However, clouds and the limitations of geostationary and polar orbiting satellites hinder the collection of high-quality thermal infrared (TIR) data. This research aims to generate hourly LST data from the Moderate Resolution Imaging Spectroradiometer (MODIS) with four daily observations. The Multi-channel Singular Spectrum Analysis (M−SSA) algorithm was used to reconstruct lost data due to clouds in the MODIS annual LST time series. Subsequently, Fourier series analysis was employed to generate hourly LST data based on the four MODIS observations per day. The developed Fourier series model was evaluated using hourly LST data from Meteosat-9 and ground surface soil temperature data at eight different Ameriflux sites. The evaluation of the Fourier series model showed that the Root Mean Square Error (RMSE) and coefficient of determination (R<sup>2</sup>) between the hourly LST data from the Meteosat-9 satellite and the hourly LST data generated by the Fourier series model using four simultaneous MODIS observations averaged 1.70 Kelvin and 0.98, respectively, throughout Iran. For Ameriflux sites, the average RMSE and R2 were 1.15 K and 0.98 between the surface soil temperature data and the surface soil temperature data generated using four simultaneous MODIS observations per day, respectively. Notably, the highest RMSE was observed during sunrise and sunset.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104341"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142902114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

BSG-WSL: BackScatter-guided weakly supervised learning for water mapping in SAR images

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2025-02-01 DOI: 10.1016/j.jag.2025.104385

Kai Wang, Zhongle Ren, Biao Hou, Weibin Li, Licheng Jiao

Extracting and analyzing water resources in Synthetic Aperture Radar (SAR) images is crucial for flood management and environmental resource planning due to the ability to monitor ground all-weather and all-time. However, extracting water entirely from high-resolution SAR images in diverse scenarios is challenging due to variable water shapes, many low-intensity land covers similar to water, and scarce labels. In this article, a BackScatter-Guided Weakly Supervised Learning (BSG-WSL) framework based on image-level labels is proposed for water extraction with the requirement of high generalization and low labeling. In BSG-WSL, a BackScatter-Guided Network (BSGNet) is proposed, where the backscatter information of water is used to guide the feature extraction process, yielding precise Class Attention Maps (CAMs) of water. Then, a morphological pseudo-label optimization algorithm is designed to employ CAMs to generate high-quality pseudo-labels. Finally, a confidence cross-entropy loss is introduced to utilize pseudo-labels to train the extraction model and achieve precise water extraction in different scenarios. Experiments on three datasets of SAR images from the GF-3 and Sentinel-1B satellites verify that the proposed method achieves state-of-the-art performance compared to other weakly supervised methods based on image-level annotations.

{"title":"BSG-WSL: BackScatter-guided weakly supervised learning for water mapping in SAR images","authors":"Kai Wang, Zhongle Ren, Biao Hou, Weibin Li, Licheng Jiao","doi":"10.1016/j.jag.2025.104385","DOIUrl":"10.1016/j.jag.2025.104385","url":null,"abstract":"<div><div>Extracting and analyzing water resources in Synthetic Aperture Radar (SAR) images is crucial for flood management and environmental resource planning due to the ability to monitor ground all-weather and all-time. However, extracting water entirely from high-resolution SAR images in diverse scenarios is challenging due to variable water shapes, many low-intensity land covers similar to water, and scarce labels. In this article, a BackScatter-Guided Weakly Supervised Learning (BSG-WSL) framework based on image-level labels is proposed for water extraction with the requirement of high generalization and low labeling. In BSG-WSL, a BackScatter-Guided Network (BSGNet) is proposed, where the backscatter information of water is used to guide the feature extraction process, yielding precise Class Attention Maps (CAMs) of water. Then, a morphological pseudo-label optimization algorithm is designed to employ CAMs to generate high-quality pseudo-labels. Finally, a confidence cross-entropy loss is introduced to utilize pseudo-labels to train the extraction model and achieve precise water extraction in different scenarios. Experiments on three datasets of SAR images from the GF-3 and Sentinel-1B satellites verify that the proposed method achieves state-of-the-art performance compared to other weakly supervised methods based on image-level annotations.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"136 ","pages":"Article 104385"},"PeriodicalIF":7.6,"publicationDate":"2025-02-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143083363","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0