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

Large-scale forest resource mapping with spatial gaps in the training data: Comparison of different modeling approaches 训练数据中存在空间缺口的大尺度森林资源制图：不同建模方法的比较

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-09 DOI: 10.1016/j.jag.2026.105104

Andras Balazs , Jukka Miettinen , Mats Nilsson , Johannes Breidenbach , Timo P. Pitkänen , Mari Myllymäki

Forest attribute maps are essential for supporting local decision-making regarding forest resource use. Such maps are produced by combining remote sensing and field data through various modeling approaches. When mapping across large areas, spatial gaps in field data used for model training are common. Our study evaluates the performance of three methods—

k

-Nearest Neighbor (

k

-NN), Random Forests (RF), and Multi-Layer Perceptron (MLP)—for forest resource mapping across Norway, Sweden, and Finland in an experimental setup with respect to availability of field data around the target area. Models were trained with sample plot sizes (

N

) ranging from 100 to 3000. RF consistently produced the most accurate predictions in terms of relative bias and RMSE. While spatial gaps in the training data (radius: 7–141 km) affected %RMSE of broad-leaved above ground biomass (AGB), they had minimal impact on %RMSE of both local and country-level predictions of total AGB and volume. For RF with

N = 3000

, %RMSE of total AGB ranged between 53%–55% in Finland and Sweden, and 70%–72% in Norway across gap sizes. However, %bias increased for local predictions across the whole study region with larger gaps: RF with

N = 500

showed bias of −12%–12% (7 km gap) and −17%–28% (78 km gap). Similarly, country-level %bias of total AGB for Norway increased from −1.7% to −3.7% with larger gaps. In conclusion, spatial gaps in training data can significantly affect bias in predictions. Therefore, forest attribute maps should always be accompanied by metadata describing the training data used.

森林属性图对于支持当地关于森林资源利用的决策至关重要。这种地图是通过各种建模方法将遥感和实地数据结合起来制作的。当绘制大区域时，用于模型训练的现场数据中的空间差距是常见的。我们的研究评估了三种方法——k-最近邻（k-NN）、随机森林（RF）和多层感知器（MLP）——在挪威、瑞典和芬兰的森林资源映射实验设置中的性能，以及目标区域周围现场数据的可用性。模型的样本量(N)为100 ~ 3000。在相对偏差和均方根误差方面，射频始终产生最准确的预测。虽然训练数据（半径为7 ~ 141 km）的空间差距影响了阔叶地上生物量（AGB）的%RMSE，但它们对地方和国家一级的总AGB和体积预测的%RMSE的影响最小。对于N=3000的RF，芬兰和瑞典总AGB的%RMSE介于53%-55%之间，挪威介于70%-72%之间。然而，在整个研究区域，局部预测的%偏差在较大的差距下增加：N=500的RF显示偏差为- 12%-12% （7 km差距）和- 17%-28% （78 km差距）。同样，挪威总AGB的国家级%偏差从- 1.7%增加到- 3.7%，差距更大。综上所述，训练数据的空间差距会显著影响预测的偏差。因此，森林属性映射应该总是伴随着描述所使用的训练数据的元数据。

{"title":"Large-scale forest resource mapping with spatial gaps in the training data: Comparison of different modeling approaches","authors":"Andras Balazs , Jukka Miettinen , Mats Nilsson , Johannes Breidenbach , Timo P. Pitkänen , Mari Myllymäki","doi":"10.1016/j.jag.2026.105104","DOIUrl":"10.1016/j.jag.2026.105104","url":null,"abstract":"<div><div>Forest attribute maps are essential for supporting local decision-making regarding forest resource use. Such maps are produced by combining remote sensing and field data through various modeling approaches. When mapping across large areas, spatial gaps in field data used for model training are common. Our study evaluates the performance of three methods—<span><math><mi>k</mi></math></span>-Nearest Neighbor (<span><math><mi>k</mi></math></span>-NN), Random Forests (RF), and Multi-Layer Perceptron (MLP)—for forest resource mapping across Norway, Sweden, and Finland in an experimental setup with respect to availability of field data around the target area. Models were trained with sample plot sizes (<span><math><mi>N</mi></math></span>) ranging from 100 to 3000. RF consistently produced the most accurate predictions in terms of relative bias and RMSE. While spatial gaps in the training data (radius: 7–141 km) affected %RMSE of broad-leaved above ground biomass (AGB), they had minimal impact on %RMSE of both local and country-level predictions of total AGB and volume. For RF with <span><math><mrow><mi>N</mi><mo>=</mo><mn>3000</mn></mrow></math></span>, %RMSE of total AGB ranged between 53%–55% in Finland and Sweden, and 70%–72% in Norway across gap sizes. However, %bias increased for local predictions across the whole study region with larger gaps: RF with <span><math><mrow><mi>N</mi><mo>=</mo><mn>500</mn></mrow></math></span> showed bias of −12%–12% (7 km gap) and −17%–28% (78 km gap). Similarly, country-level %bias of total AGB for Norway increased from −1.7% to −3.7% with larger gaps. In conclusion, spatial gaps in training data can significantly affect bias in predictions. Therefore, forest attribute maps should always be accompanied by metadata describing the training data used.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105104"},"PeriodicalIF":8.6,"publicationDate":"2026-01-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925602","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}

引用次数: 0

A novel two-step method for ratoon rice mapping using Sentinel-1/2 time series 基于Sentinel-1/2时间序列的两步水稻制图方法

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-08 DOI: 10.1016/j.jag.2025.105083

Yue Wang , Yuechen Li , Xiaolin Zhu , Jin Chen , Ruyin Cao , Xiong Yao , Wujun Zhang

Ratoon rice plays a vital role in boosting land productivity and contributing to stable food supplies under the context of global climate change. This system offers these advantages by demonstrating the capacity to produce an additional grain yield of 5–6 t ha⁻¹ from the ratoon season, while simultaneously reducing the growth duration by 44–48 days compared to conventional double-season rice cultivation. However, its precise spatial distribution remains unclear under varying climatic and surface conditions, and remote sensing-based monitoring of ratoon rice has received limited attention. Existing methods often struggle to accurately distinguish ratoon rice from other morphologically similar types, such as double-season rice, and are further hampered by frequent cloud cover and rainfall, compromising optical sensing effectiveness. This study proposes a two-step ratoon rice (TSRR) mapping method using multi-source remote sensing data at the parcel scale. The TSRR method integrates Sentinel-1A synthetic aperture radar (SAR) and Sentinel-2 optical imagery, utilizing the distinct growth characteristics of main-season and ratoon rice. It employs object-based segmentation and two novel indices—the SAR-based paddy rice index (SPRI) and the SAR-based ratoon rice index (SRRI)—without relying on detailed phenological information. Results indicate that the TSRR method effectively distinguishes ratoon rice from other paddy rice types, achieving an average overall accuracy (OA) of 0.87, with particularly high performance in separating ratoon rice from double-season rice. The TSRR method demonstrates strong robustness and transferability across different regions, which can provide a reliable solution for large-scale paddy rice mapping, especially in cloud-prone areas with limited optical data availability, and offers valuable support for crop monitoring, yield estimation, and national agricultural inventory initiatives.

在全球气候变化的背景下，大米在提高土地生产力和稳定粮食供应方面发挥着至关重要的作用。该系统具有这些优势，因为它显示了从再生季节开始额外生产5-6吨/公顷粮食的能力，同时与传统的双季稻种植相比，生长期缩短了44-48天。然而，在不同的气候和地表条件下，其精确的空间分布尚不清楚，基于遥感的水稻监测受到的关注有限。现有的方法往往难以准确区分水稻与其他形态相似的水稻，如双季稻，并且由于频繁的云层和降雨，进一步阻碍了光学传感的有效性。本文提出了一种基于多源遥感数据的两步成片水稻（TSRR）制图方法。TSRR方法结合了Sentinel-1A合成孔径雷达（SAR）和Sentinel-2光学图像，利用了主季稻和次季稻不同的生长特征。该方法采用了基于目标的分割和两个新的指数——基于sar的水稻指数（SPRI）和基于sar的生长期水稻指数（SRRI），而不依赖于详细的物候信息。结果表明，TSRR方法能有效区分籼稻和其他水稻类型，平均总体准确率（OA）为0.87，其中在区分籼稻和双季稻方面表现尤为优异。TSRR方法具有较强的鲁棒性和跨区域可转移性，可为大规模水稻制图提供可靠的解决方案，特别是在光学数据可用性有限的多云地区，并为作物监测、产量估计和国家农业清查举措提供有价值的支持。

{"title":"A novel two-step method for ratoon rice mapping using Sentinel-1/2 time series","authors":"Yue Wang , Yuechen Li , Xiaolin Zhu , Jin Chen , Ruyin Cao , Xiong Yao , Wujun Zhang","doi":"10.1016/j.jag.2025.105083","DOIUrl":"10.1016/j.jag.2025.105083","url":null,"abstract":"<div><div>Ratoon rice plays a vital role in boosting land productivity and contributing to stable food supplies under the context of global climate change. This system offers these advantages by demonstrating the capacity to produce an additional grain yield of 5–6 t ha<sup>−1</sup> from the ratoon season, while simultaneously reducing the growth duration by 44–48 days compared to conventional double-season rice cultivation. However, its precise spatial distribution remains unclear under varying climatic and surface conditions, and remote sensing-based monitoring of ratoon rice has received limited attention. Existing methods often struggle to accurately distinguish ratoon rice from other morphologically similar types, such as double-season rice, and are further hampered by frequent cloud cover and rainfall, compromising optical sensing effectiveness. This study proposes a two-step ratoon rice (TSRR) mapping method using multi-source remote sensing data at the parcel scale. The TSRR method integrates Sentinel-1A synthetic aperture radar (SAR) and Sentinel-2 optical imagery, utilizing the distinct growth characteristics of main-season and ratoon rice. It employs object-based segmentation and two novel indices—the SAR-based paddy rice index (SPRI) and the SAR-based ratoon rice index (SRRI)—without relying on detailed phenological information. Results indicate that the TSRR method effectively distinguishes ratoon rice from other paddy rice types, achieving an average overall accuracy (OA) of 0.87, with particularly high performance in separating ratoon rice from double-season rice. The TSRR method demonstrates strong robustness and transferability across different regions, which can provide a reliable solution for large-scale paddy rice mapping, especially in cloud-prone areas with limited optical data availability, and offers valuable support for crop monitoring, yield estimation, and national agricultural inventory initiatives.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105083"},"PeriodicalIF":8.6,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926334","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}

引用次数: 0

Modular and adaptive implementation of Semantic Segmentation Models for Satellite Images and Open Source tools suitable for complex geographical contexts 卫星图像语义分割模型的模块化和自适应实现以及适用于复杂地理环境的开源工具

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-08 DOI: 10.1016/j.jag.2025.105069

Adrien Le Guillou , Simona Niculescu

Semantic segmentation, the process of assigning a semantic label to each pixel in an image, is a critical computer vision task for extracting detailed information from remote sensing data. However, its application to complex geographical contexts, such as coastal wetlands, is often constrained by the need for highly specialized implementations, class imbalance, and limited accessibility for non-specialists. This paper introduces a novel, modular, and adaptive open-source framework for semantic segmentation tailored to satellite imagery. Designed for maximum flexibility, the framework supports both binary and multi-class segmentation tasks and incorporates specific training strategies to handle severe class imbalances inherent in ecological detection, such as salt marsh mapping. The implementation provides a fully configurable pipeline that bridges the gap between Geographic Information Systems (GIS) and Deep Learning (DL). It integrates QGIS for intuitive spatial preprocessing and grid generation with a Python-based training and prediction workflow, thereby democratizing access to advanced segmentation techniques. The framework is architecture-agnostic, allowing the seamless deployment and benchmarking of various state-of-the-art encoder–decoder models, which are effective at combining multi-scale contextual information with high spatial resolution. A key contribution is the integration of a multifaceted training methodology that includes hybrid loss functions with dynamic class weighting and spectral-consistent data augmentation to ensure robust model generalization from limited and imbalanced datasets. We demonstrate the framework’s efficacy and scalability through two distinct case studies: a multi-class land cover classification on the Crozon Peninsula using Pléiades and a binary salt marsh detection in the Mont-Saint-Michel Bay Sentinel-2 imagery. The results show that accurate segmentation can be achieved with modest computational resources, promoting more sustainable and ethical AI applications in environmental monitoring. This work provides a critical tool for researchers and practitioners aiming to apply advanced DL segmentation to domain specific remote sensing challenges beyond conventional benchmarks.

语义分割是为图像中的每个像素分配语义标签的过程，是从遥感数据中提取详细信息的关键计算机视觉任务。然而，它在复杂地理环境中的应用，如沿海湿地，往往受到高度专业化实施的需要、类别不平衡和非专业人员的有限可及性的限制。本文介绍了一种新颖的、模块化的、自适应的开源框架，用于卫星图像的语义分割。该框架具有最大的灵活性，支持二元和多类分割任务，并结合特定的训练策略来处理生态检测中固有的严重类不平衡，例如盐沼测绘。该实现提供了一个完全可配置的管道，弥合了地理信息系统（GIS）和深度学习（DL）之间的差距。它将QGIS与基于python的训练和预测工作流程集成在一起，用于直观的空间预处理和网格生成，从而使高级分割技术的访问民主化。该框架与架构无关，允许各种最先进的编码器-解码器模型的无缝部署和基准测试，这些模型可以有效地将多尺度上下文信息与高空间分辨率相结合。一个关键的贡献是集成了多方面的训练方法，包括混合损失函数与动态类加权和频谱一致的数据增强，以确保从有限和不平衡的数据集稳健的模型泛化。我们通过两个不同的案例研究证明了该框架的有效性和可扩展性：在Crozon半岛使用placimiades进行多类土地覆盖分类，以及在mont saint - michel湾Sentinel-2图像中进行二元盐沼检测。结果表明，在适度的计算资源下，可以实现准确的分割，促进人工智能在环境监测中的应用更具可持续性和伦理性。这项工作为研究人员和实践者提供了一个重要的工具，旨在将先进的深度学习分割应用于超越传统基准的特定领域的遥感挑战。

{"title":"Modular and adaptive implementation of Semantic Segmentation Models for Satellite Images and Open Source tools suitable for complex geographical contexts","authors":"Adrien Le Guillou , Simona Niculescu","doi":"10.1016/j.jag.2025.105069","DOIUrl":"10.1016/j.jag.2025.105069","url":null,"abstract":"<div><div>Semantic segmentation, the process of assigning a semantic label to each pixel in an image, is a critical computer vision task for extracting detailed information from remote sensing data. However, its application to complex geographical contexts, such as coastal wetlands, is often constrained by the need for highly specialized implementations, class imbalance, and limited accessibility for non-specialists. This paper introduces a novel, modular, and adaptive open-source framework for semantic segmentation tailored to satellite imagery. Designed for maximum flexibility, the framework supports both binary and multi-class segmentation tasks and incorporates specific training strategies to handle severe class imbalances inherent in ecological detection, such as salt marsh mapping. The implementation provides a fully configurable pipeline that bridges the gap between Geographic Information Systems (GIS) and Deep Learning (DL). It integrates QGIS for intuitive spatial preprocessing and grid generation with a Python-based training and prediction workflow, thereby democratizing access to advanced segmentation techniques. The framework is architecture-agnostic, allowing the seamless deployment and benchmarking of various state-of-the-art encoder–decoder models, which are effective at combining multi-scale contextual information with high spatial resolution. A key contribution is the integration of a multifaceted training methodology that includes hybrid loss functions with dynamic class weighting and spectral-consistent data augmentation to ensure robust model generalization from limited and imbalanced datasets. We demonstrate the framework’s efficacy and scalability through two distinct case studies: a multi-class land cover classification on the Crozon Peninsula using Pléiades and a binary salt marsh detection in the Mont-Saint-Michel Bay Sentinel-2 imagery. The results show that accurate segmentation can be achieved with modest computational resources, promoting more sustainable and ethical AI applications in environmental monitoring. This work provides a critical tool for researchers and practitioners aiming to apply advanced DL segmentation to domain specific remote sensing challenges beyond conventional benchmarks.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105069"},"PeriodicalIF":8.6,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926335","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}

引用次数: 0

Integrated and simultaneous mapping of blue carbon ecosystems by using tide-level, phenological, and biophysical features from optical and SAR images 利用来自光学和SAR图像的潮位、物候和生物物理特征对蓝碳生态系统进行综合和同步制图

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-08 DOI: 10.1016/j.jag.2025.105076

Leping Wang , Qian Zhang , Yangfan Li

Blue carbon ecosystems (BCEs) are nature-based solutions critical for mitigating climate change and biodiversity loss. Accurate mapping of BCEs is fundamental to carbon accounting, maximizing their ecosystem service value, and informing conservation and restoration efforts. Yet, most existing studies focus on single ecosystem mapping and lack multi-class classification approaches capable of addressing the spectral similarity among different BCEs. To address this issue, we developed a novel algorithm on Google Earth Engine, namely Multi-class Blue Carbon Ecosystem Mapping by integrating Tide-level, Phenological, and Biophysical features (MBCEM-TPB) to simultaneously map mangroves, saltmarshes, and intertidal seagrass meadows, thereby characterizing the full composition of BCEs. Specifically, we first composited multi-temporal imagery under different tidal levels, phenological stages, and biophysical features from Sentinel-1 and Sentinel-2 data. Based on spectral similarity principles, we performed training sample migration and then generated interannual blue carbon maps for 2019, 2021, and 2023 using Random Forest classifier. The algorithm was evaluated across eight study sites encompassing different BCEs combinations (two or three ecosystem types) spanning diverse climate zones, bioregions, and levels of ecosystem complexity. The overall accuracy of the MBCEM-TPB algorithm exceeded 93.65% across three periods, demonstrating its robustness and generalizability, even in complex intertidal landscapes. This study provides the first unified multi-class classification algorithm for BCEs and offers a generalizable approach applicable at global scales, supporting refined blue carbon accounting and ecosystem management.

蓝碳生态系统（bce）是基于自然的解决方案，对减缓气候变化和生物多样性丧失至关重要。bce的精确测绘是碳核算、最大化其生态系统服务价值以及为保护和恢复工作提供信息的基础。然而，现有的研究大多集中在单一生态系统的制图上，缺乏能够解决不同生物多样性之间光谱相似性的多类分类方法。为了解决这一问题，我们在谷歌Earth Engine上开发了一种新的算法，即结合潮位、物质性和生物物理特征的Multi-class Blue Carbon Ecosystem Mapping (MBCEM-TPB)，同时绘制红树林、盐沼和潮间带海草草甸的蓝碳生态系统图，从而表征了潮间带海草草甸的全部组成。具体来说，我们首先合成了Sentinel-1和Sentinel-2数据在不同潮位、物候阶段和生物物理特征下的多时相图像。基于光谱相似原理，我们进行了训练样本迁移，然后使用随机森林分类器生成了2019年、2021年和2023年的年际蓝碳图。该算法在8个研究地点进行了评估，这些研究地点涵盖了不同的bce组合（两种或三种生态系统类型），跨越了不同的气候带、生物区和生态系统复杂性水平。MBCEM-TPB算法在三个周期内的总体精度超过93.65%，即使在复杂的潮间带景观中也显示出其鲁棒性和泛化性。该研究提供了首个统一的bce多类分类算法，并提供了一种适用于全球尺度的可推广方法，为精细化的蓝碳核算和生态系统管理提供支持。

{"title":"Integrated and simultaneous mapping of blue carbon ecosystems by using tide-level, phenological, and biophysical features from optical and SAR images","authors":"Leping Wang , Qian Zhang , Yangfan Li","doi":"10.1016/j.jag.2025.105076","DOIUrl":"10.1016/j.jag.2025.105076","url":null,"abstract":"<div><div>Blue carbon ecosystems (BCEs) are nature-based solutions critical for mitigating climate change and biodiversity loss. Accurate mapping of BCEs is fundamental to carbon accounting, maximizing their ecosystem service value, and informing conservation and restoration efforts. Yet, most existing studies focus on single ecosystem mapping and lack multi-class classification approaches capable of addressing the spectral similarity among different BCEs. To address this issue, we developed a novel algorithm on Google Earth Engine, namely Multi-class Blue Carbon Ecosystem Mapping by integrating Tide-level, Phenological, and Biophysical features (MBCEM-TPB) to simultaneously map mangroves, saltmarshes, and intertidal seagrass meadows, thereby characterizing the full composition of BCEs. Specifically, we first composited multi-temporal imagery under different tidal levels, phenological stages, and biophysical features from Sentinel-1 and Sentinel-2 data. Based on spectral similarity principles, we performed training sample migration and then generated interannual blue carbon maps for 2019, 2021, and 2023 using Random Forest classifier. The algorithm was evaluated across eight study sites encompassing different BCEs combinations (two or three ecosystem types) spanning diverse climate zones, bioregions, and levels of ecosystem complexity. The overall accuracy of the MBCEM-TPB algorithm exceeded 93.65% across three periods, demonstrating its robustness and generalizability, even in complex intertidal landscapes. This study provides the first unified multi-class classification algorithm for BCEs and offers a generalizable approach applicable at global scales, supporting refined blue carbon accounting and ecosystem management.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105076"},"PeriodicalIF":8.6,"publicationDate":"2026-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925667","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}

引用次数: 0

Integrating linear and circular polarization features for PolSAR land cover classification with deep learning 基于深度学习的线性和圆极化特征融合PolSAR土地覆盖分类

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2026.105090

Shuaiying Zhang , Zhen Dong , Huadong Lin , Zhendong Zhang , Jinran Wu , Sinong Quan , Wentao An , Tong Li , Rajiv Pandey

Existing deep learning methods for ecological monitoring using polarimetric synthetic aperture radar (PolSAR) imagery primarily rely on coherency (

T

) or covariance (

C

) matrices derived from the linear polarization basis, often overlooking scattering information inherent in alternative polarization representations. To address this limitation, this study proposes a novel classification framework that explicitly incorporates a circular polarization basis into the PolSAR deep learning workflow. A circular coherency matrix (

Cir

), analogous to the conventional

T

matrix, was first derived through polarization basis transformation. Subsequently, a multi-basis input scheme was introduced to fuse linear and circular polarization features to enhance feature representation and information utilization. The proposed framework was validated on two benchmark datasets using multiple deep learning models, achieving state-of-the-art classification accuracies of 97.70% and 98.58%.Compared with standard linear-basis approaches, the proposed scheme yielded accuracy improvements of 2.86% over the

T

-matrix-based method and 2.26% over the

C

-matrix-based method. In addition, the incorporation of circular polarization features significantly enhanced physical interpretability, particularly for structurally complex targets such as forests and buildings. Overall, the findings provide an effective technical pathway for intelligent land cover classification and broader ecological monitoring. The source code and datasets are available at https://github.com/zhangssy/Circular-Polarization-Basis-Implementation.

利用偏振合成孔径雷达（PolSAR）图像进行生态监测的现有深度学习方法主要依赖于线性偏振基衍生的相干(T)或协方差(C)矩阵，往往忽略了替代偏振表示中固有的散射信息。为了解决这一限制，本研究提出了一种新的分类框架，该框架明确地将圆极化基础纳入PolSAR深度学习工作流程。首先通过极化基变换导出了与传统T矩阵类似的圆相干矩阵（Cir）。随后，引入多基输入方案融合线极化和圆极化特征，增强特征表示和信息利用率。使用多个深度学习模型在两个基准数据集上验证了所提出的框架，达到了97.70%和98.58%的最先进分类准确率。与标准线性基方法相比，该方法的准确率比基于t矩阵的方法提高2.86%，比基于c矩阵的方法提高2.26%。此外，圆偏振特征的结合显著提高了物理可解释性，特别是对于结构复杂的目标，如森林和建筑物。研究结果为智能土地覆盖分类和更广泛的生态监测提供了有效的技术途径。源代码和数据集可从https://github.com/zhangssy/Circular-Polarization-Basis-Implementation获得。

{"title":"Integrating linear and circular polarization features for PolSAR land cover classification with deep learning","authors":"Shuaiying Zhang , Zhen Dong , Huadong Lin , Zhendong Zhang , Jinran Wu , Sinong Quan , Wentao An , Tong Li , Rajiv Pandey","doi":"10.1016/j.jag.2026.105090","DOIUrl":"10.1016/j.jag.2026.105090","url":null,"abstract":"<div><div>Existing deep learning methods for ecological monitoring using polarimetric synthetic aperture radar (PolSAR) imagery primarily rely on coherency (<span><math><mi>T</mi></math></span>) or covariance (<span><math><mi>C</mi></math></span>) matrices derived from the linear polarization basis, often overlooking scattering information inherent in alternative polarization representations. To address this limitation, this study proposes a novel classification framework that explicitly incorporates a circular polarization basis into the PolSAR deep learning workflow. A circular coherency matrix (<span><math><mi>Cir</mi></math></span>), analogous to the conventional <span><math><mi>T</mi></math></span> matrix, was first derived through polarization basis transformation. Subsequently, a multi-basis input scheme was introduced to fuse linear and circular polarization features to enhance feature representation and information utilization. The proposed framework was validated on two benchmark datasets using multiple deep learning models, achieving state-of-the-art classification accuracies of 97.70% and 98.58%.Compared with standard linear-basis approaches, the proposed scheme yielded accuracy improvements of 2.86% over the <span><math><mi>T</mi></math></span>-matrix-based method and 2.26% over the <span><math><mi>C</mi></math></span>-matrix-based method. In addition, the incorporation of circular polarization features significantly enhanced physical interpretability, particularly for structurally complex targets such as forests and buildings. Overall, the findings provide an effective technical pathway for intelligent land cover classification and broader ecological monitoring. The source code and datasets are available at <span><span>https://github.com/zhangssy/Circular-Polarization-Basis-Implementation</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 105090"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926336","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}

引用次数: 0

Dense 3D displacement estimation for landslide monitoring via fusion of TLS point clouds and embedded RGB images 基于TLS点云和嵌入式RGB图像融合的滑坡监测密集三维位移估计

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2026.105093

Zhaoyi Wang , Jemil Avers Butt , Shengyu Huang , Tomislav Medić , Andreas Wieser

Landslide monitoring is essential for understanding geohazards and mitigating associated risks. Existing point cloud-based methods, however, typically rely on either geometric or radiometric information and often yield sparse or non-3D displacement estimates. In this paper, we propose a hierarchical partitioning-based coarse-to-fine approach that fuses 3D point clouds and co-registered RGB images to estimate dense 3D displacement vector fields. Patch-level matches are constructed using both 3D geometry and 2D image features, refined via geometric consistency checks, and followed by rigid transformation estimation per match. Experimental results on two real-world landslide datasets demonstrate that the proposed method produces 3D displacement estimates with high spatial coverage (79% and 97%) and accuracy. Deviations in displacement magnitude with respect to external measurements (total station or GNSS observations) are 0.15 m and 0.25 m on the two datasets, respectively, and only 0.07 m and 0.20 m compared to manually derived references, all below the mean scan resolutions (0.08 m and 0.30 m). Compared with the state-of-the-art method F2S3, the proposed approach improves spatial coverage while maintaining comparable accuracy. The proposed approach offers a practical and adaptable solution for TLS-based landslide monitoring and is extensible to other types of point clouds and monitoring tasks.

滑坡监测对于了解地质灾害和减轻相关风险至关重要。然而，现有的基于点云的方法通常依赖于几何或辐射信息，并且经常产生稀疏或非三维位移估计。在本文中，我们提出了一种基于分层划分的粗到精方法，该方法融合了三维点云和共配准的RGB图像来估计密集的三维位移向量场。使用3D几何和2D图像特征构建补丁级匹配，通过几何一致性检查进行细化，然后对每个匹配进行刚性变换估计。在两个实际滑坡数据集上的实验结果表明，该方法产生的三维位移估计具有较高的空间覆盖率（79%和97%）和精度。相对于外部测量（全站站或GNSS观测），两个数据集的位移量级偏差分别为0.15 m和0.25 m，与手动导出的参考数据相比，偏差仅为0.07 m和0.20 m，均低于平均扫描分辨率（0.08 m和0.30 m）。与最先进的方法F2S3相比，该方法在保持相当精度的同时提高了空间覆盖。该方法为基于tls的滑坡监测提供了一种实用且适应性强的解决方案，并可扩展到其他类型的点云和监测任务中。

{"title":"Dense 3D displacement estimation for landslide monitoring via fusion of TLS point clouds and embedded RGB images","authors":"Zhaoyi Wang , Jemil Avers Butt , Shengyu Huang , Tomislav Medić , Andreas Wieser","doi":"10.1016/j.jag.2026.105093","DOIUrl":"10.1016/j.jag.2026.105093","url":null,"abstract":"<div><div>Landslide monitoring is essential for understanding geohazards and mitigating associated risks. Existing point cloud-based methods, however, typically rely on either geometric or radiometric information and often yield sparse or non-3D displacement estimates. In this paper, we propose a hierarchical partitioning-based coarse-to-fine approach that fuses 3D point clouds and co-registered RGB images to estimate dense 3D displacement vector fields. Patch-level matches are constructed using both 3D geometry and 2D image features, refined via geometric consistency checks, and followed by rigid transformation estimation per match. Experimental results on two real-world landslide datasets demonstrate that the proposed method produces 3D displacement estimates with high spatial coverage (79% and 97%) and accuracy. Deviations in displacement magnitude with respect to external measurements (total station or GNSS observations) are 0.15 m and 0.25 m on the two datasets, respectively, and only 0.07 m and 0.20 m compared to manually derived references, all below the mean scan resolutions (0.08 m and 0.30 m). Compared with the state-of-the-art method F2S3, the proposed approach improves spatial coverage while maintaining comparable accuracy. The proposed approach offers a practical and adaptable solution for TLS-based landslide monitoring and is extensible to other types of point clouds and monitoring tasks.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105093"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926452","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}

引用次数: 0

Causative environment-informed active landslide detection with InSAR monitoring: Empowering reservoir slope hazard management 基于InSAR监测的诱发环境信息主动滑坡探测：增强水库边坡灾害管理能力

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2025.105079

Shuhao Ran , Wei Zhou , Yonghong Weng , Fudong Chi , Di Wang , Xuexing Cao , Gang Ma

The accurate detection of active landslides influenced by reservoir water-level fluctuations is critically important for safeguarding hydropower infrastructure and protecting lives and property along riverbanks. However, most current InSAR-based approaches primarily rely on abrupt changes in phase or velocity gradients, while largely ignoring the causative environmental conditions that control landslide initiation. This study proposes a novel method for causative environment-informed active landslide detection with InSAR monitoring in the reservoir area. First, a four-directional velocity gradient matrix is constructed from time-series deformation data. Spatial morphology analysis is then used to filter out noise, extracting active deformation areas (ADAs) with continuous boundaries. Subsequently, five slope unit-based landslide susceptibility prediction (LSP) models are constructed to learn the causative environmental conditions of reservoir landslides systematically. Finally, an area under the curve (AUC) weighted voting strategy is employed to integrate deformation signals with environmental constraints, enabling the reliable identification of active landslides. Applied to the DHQ and HD reservoir areas of the Lancang River basin, the method identified 89 ADAs, achieving a 96% spatial overlap with manually interpreted deformation areas. Under the constraints of causative environmental factors, the overall F₁-Score for identifying active landslides in the reservoir area reached 78.15%. Model interpretability analysis revealed that distance to river (DtR) plays a dominant role in both the formation and detection of reservoir landslides. The study further identifies key landslide-causing environment factors (LEFs) and high-susceptibility areas, providing valuable data support and decision-making guidance for reservoir slope disaster management, landslide monitoring, early warning, and risk prevention.

准确探测受水库水位波动影响的活动性滑坡，对于保障水电基础设施和保护河岸生命财产安全至关重要。然而，目前大多数基于insar的方法主要依赖于相位或速度梯度的突变，而在很大程度上忽略了控制滑坡发生的环境条件。本研究提出了一种基于InSAR监测的库区成因环境主动滑坡检测新方法。首先，利用时间序列变形数据构造四向速度梯度矩阵；然后利用空间形态分析滤除噪声，提取具有连续边界的活动变形区（ADAs）。在此基础上，构建了5个基于边坡单元的滑坡敏感性预测（LSP）模型，系统地了解了水库滑坡的成因环境条件。最后，采用曲线下面积（area under the curve， AUC）加权投票策略，将变形信号与环境约束相结合，实现对活动滑坡的可靠识别。将该方法应用于澜沧江流域的DHQ和HD库区，识别出89个ADAs，与人工解释的变形区有96%的空间重叠。在成因环境因素约束下，库区活动性滑坡识别总分总分为78.15%。模型可解释性分析表明，距河距离（DtR）在水库滑坡的形成和探测中都起着主导作用。研究进一步确定了诱发滑坡的关键环境因子（LEFs）和高易感性区域，为水库边坡灾害管理、滑坡监测、预警和风险防范提供了有价值的数据支持和决策指导。

{"title":"Causative environment-informed active landslide detection with InSAR monitoring: Empowering reservoir slope hazard management","authors":"Shuhao Ran , Wei Zhou , Yonghong Weng , Fudong Chi , Di Wang , Xuexing Cao , Gang Ma","doi":"10.1016/j.jag.2025.105079","DOIUrl":"10.1016/j.jag.2025.105079","url":null,"abstract":"<div><div>The accurate detection of active landslides influenced by reservoir water-level fluctuations is critically important for safeguarding hydropower infrastructure and protecting lives and property along riverbanks. However, most current InSAR-based approaches primarily rely on abrupt changes in phase or velocity gradients, while largely ignoring the causative environmental conditions that control landslide initiation. This study proposes a novel method for causative environment-informed active landslide detection with InSAR monitoring in the reservoir area. First, a four-directional velocity gradient matrix is constructed from time-series deformation data. Spatial morphology analysis is then used to filter out noise, extracting active deformation areas (ADAs) with continuous boundaries. Subsequently, five slope unit-based landslide susceptibility prediction (LSP) models are constructed to learn the causative environmental conditions of reservoir landslides systematically. Finally, an area under the curve (AUC) weighted voting strategy is employed to integrate deformation signals with environmental constraints, enabling the reliable identification of active landslides. Applied to the DHQ and HD reservoir areas of the Lancang River basin, the method identified 89 ADAs, achieving a 96% spatial overlap with manually interpreted deformation areas. Under the constraints of causative environmental factors, the overall F<sub>1</sub>-Score for identifying active landslides in the reservoir area reached 78.15%. Model interpretability analysis revealed that distance to river (DtR) plays a dominant role in both the formation and detection of reservoir landslides. The study further identifies key landslide-causing environment factors (LEFs) and high-susceptibility areas, providing valuable data support and decision-making guidance for reservoir slope disaster management, landslide monitoring, early warning, and risk prevention.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105079"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926453","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}

引用次数: 0

Global elevational shifts and drivers of alpine treelines 高山树线的全球海拔变化和驱动因素

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2026.105088

Tianchen Liang , Feng Tian , Linqing Zou , Mathieu Gravey , Sabine B. Rumpf

Treelines are generally expected to respond to climate warming by shifting to higher elevations. However, significant variations in treeline shifts across different mountain ranges have been observed, and the underlying reasons for these differences remain unclear due to the lack of comprehensive global treeline mapping. This study provides a global analysis of treeline shifts using satellite-derived forest canopy height data at a 30-meter resolution, combined with a model of potential treeline positions based on climatic data. The elevations of both observed and potential treelines were mapped for the years 2000 and 2020, and the elevational shifts in treeline positions were quantified over this period. On a global scale, 42% of observed treelines shifted upslope and 25% downslope. However, these shifts generally lagged behind climate-driven potential treeline elevations. By contrast, the rate of treeline shifts was strongly influenced by anthropogenic disturbances. Fire disturbances contributed to 38% of downslope shifts, and regions with lower overall anthropogenic disturbance had faster upslope shifts of observed treelines and thus consequently smaller lags. These findings emphasize the importance of considering both climatic and anthropogenic factors when assessing treeline dynamics and suggest that historical and ongoing human impacts play a critical role in modulating treeline responses to climate change.

人们普遍认为，树线会通过向更高海拔地区迁移来应对气候变暖。然而，已经观察到不同山脉的树线变化存在显著差异，由于缺乏全面的全球树线制图，造成这些差异的根本原因尚不清楚。本研究利用卫星获得的30米分辨率的森林冠层高度数据，结合基于气候数据的潜在树线位置模型，对树线变化进行了全球分析。绘制了2000年和2020年观测树线和潜在树线的高程图，并量化了这一时期树线位置的高程变化。在全球范围内，42%的观测树线向上移动，25%向下移动。然而，这些变化通常落后于气候驱动的潜在树木线高度。相比之下，树线移动的速度受到人为干扰的强烈影响。火灾干扰贡献了38%的下坡移动，总体人为干扰较小的地区观测到的树线上坡移动更快，因此滞后较小。这些发现强调了在评估树线动态时同时考虑气候和人为因素的重要性，并表明历史和持续的人类影响在调节树线对气候变化的响应方面发挥着关键作用。

{"title":"Global elevational shifts and drivers of alpine treelines","authors":"Tianchen Liang , Feng Tian , Linqing Zou , Mathieu Gravey , Sabine B. Rumpf","doi":"10.1016/j.jag.2026.105088","DOIUrl":"10.1016/j.jag.2026.105088","url":null,"abstract":"<div><div>Treelines are generally expected to respond to climate warming by shifting to higher elevations. However, significant variations in treeline shifts across different mountain ranges have been observed, and the underlying reasons for these differences remain unclear due to the lack of comprehensive global treeline mapping. This study provides a global analysis of treeline shifts using satellite-derived forest canopy height data at a 30-meter resolution, combined with a model of potential treeline positions based on climatic data. The elevations of both observed and potential treelines were mapped for the years 2000 and 2020, and the elevational shifts in treeline positions were quantified over this period. On a global scale, 42% of observed treelines shifted upslope and 25% downslope. However, these shifts generally lagged behind climate-driven potential treeline elevations. By contrast, the rate of treeline shifts was strongly influenced by anthropogenic disturbances. Fire disturbances contributed to 38% of downslope shifts, and regions with lower overall anthropogenic disturbance had faster upslope shifts of observed treelines and thus consequently smaller lags. These findings emphasize the importance of considering both climatic and anthropogenic factors when assessing treeline dynamics and suggest that historical and ongoing human impacts play a critical role in modulating treeline responses to climate change.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105088"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926451","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}

引用次数: 0

Construction and visualization analysis of urban night multispectral inversion model based on SDGSAT-1 glimmer imagery 基于SDGSAT-1微光影像的城市夜间多光谱反演模型构建及可视化分析

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2025.105065

Ming Liu , Ruicong Li , Lie Feng , Ezzaddeen Ali Mohammed Saeed AL-Mowallad , Weili Jiao , Han Zhang , Fei Xu

The spectrum is a key physical quantity characterizing the urban nighttime light environment. However, due to the prevalent use of single-band observations in conventional nighttime light remote sensing, studies on urban nighttime spectral characteristics remain relatively weak. With the advancement of multispectral remote sensing, band-wise spectral retrieval has become feasible. In this study, based on SDGSAT-1 nighttime multispectral imagery and ground-based measurements, we compare the spectral ranges of human visual perception and satellite sensors, retrieve the urban nighttime RGB-band spectral distribution, and construct a light-environment inversion map and a blue-light ratio map. The results show that: (1) configuring bands with reference to the human visual spectral range is more suitable for remote sensing spectral retrieval, among which the B band exhibits the highest correlation with ground observations (correlation coefficient 0.879), followed by the R and G bands (0.700 and 0.688, respectively). (2) A comparison of six linear regression models with three machine learning models—random forest, back-propagation (BP) neural network, and support vector regression—indicates that machine learning models overall outperform linear models, while the three machine learning approaches achieve comparable accuracies, with cross-validated R² values of approximately 0.65–0.70. (3) Considering residual characteristics and uncertainty analysis, the random forest model is selected as the primary inversion model to retrieve the nighttime spectra of the main urban area of Dalian. The band-wise and blue-light ratio maps reveal the spatial patterns of high-luminance functional areas such as traffic corridors, commercial districts, and landscape lighting, demonstrating that multispectral nighttime light remote sensing can provide important support for urban lighting planning and sustainable urban development.

光谱是表征城市夜间光环境的关键物理量。然而，由于传统夜光遥感普遍采用单波段观测，对城市夜间光谱特征的研究相对薄弱。随着多光谱遥感技术的发展，波段光谱检索已成为可能。本研究基于SDGSAT-1夜间多光谱影像和地面实测数据，对比人类视觉感知和卫星传感器的光谱范围，检索城市夜间rgb波段光谱分布，构建光环境反演图和蓝光比图。结果表明：(1)参考人类视觉光谱范围配置波段更适合遥感光谱检索，其中B波段与地面观测相关性最高（相关系数为0.879），其次是R波段和G波段（相关系数分别为0.700和0.688）。(2)六种线性回归模型与三种机器学习模型（随机森林、BP神经网络和支持向量回归）的比较表明，机器学习模型总体上优于线性模型，而三种机器学习方法的精度相当，交叉验证的R2值约为0.65-0.70。(3)考虑残差特征和不确定性分析，选择随机森林模型作为反演大连市主城区夜间光谱的主要模型。波段图和蓝光比图揭示了交通走廊、商业区和景观照明等高亮度功能区的空间格局，表明多光谱夜间灯光遥感可以为城市照明规划和城市可持续发展提供重要支持。

{"title":"Construction and visualization analysis of urban night multispectral inversion model based on SDGSAT-1 glimmer imagery","authors":"Ming Liu , Ruicong Li , Lie Feng , Ezzaddeen Ali Mohammed Saeed AL-Mowallad , Weili Jiao , Han Zhang , Fei Xu","doi":"10.1016/j.jag.2025.105065","DOIUrl":"10.1016/j.jag.2025.105065","url":null,"abstract":"<div><div>The spectrum is a key physical quantity characterizing the urban nighttime light environment. However, due to the prevalent use of single-band observations in conventional nighttime light remote sensing, studies on urban nighttime spectral characteristics remain relatively weak. With the advancement of multispectral remote sensing, band-wise spectral retrieval has become feasible. In this study, based on SDGSAT-1 nighttime multispectral imagery and ground-based measurements, we compare the spectral ranges of human visual perception and satellite sensors, retrieve the urban nighttime RGB-band spectral distribution, and construct a light-environment inversion map and a blue-light ratio map. The results show that: (1) configuring bands with reference to the human visual spectral range is more suitable for remote sensing spectral retrieval, among which the B band exhibits the highest correlation with ground observations (correlation coefficient 0.879), followed by the R and G bands (0.700 and 0.688, respectively). (2) A comparison of six linear regression models with three machine learning models—random forest, back-propagation (BP) neural network, and support vector regression—indicates that machine learning models overall outperform linear models, while the three machine learning approaches achieve comparable accuracies, with cross-validated R<sup>2</sup> values of approximately 0.65–0.70. (3) Considering residual characteristics and uncertainty analysis, the random forest model is selected as the primary inversion model to retrieve the nighttime spectra of the main urban area of Dalian. The band-wise and blue-light ratio maps reveal the spatial patterns of high-luminance functional areas such as traffic corridors, commercial districts, and landscape lighting, demonstrating that multispectral nighttime light remote sensing can provide important support for urban lighting planning and sustainable urban development.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105065"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925666","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}

引用次数: 0

Multi-perspective occlusion-free urban navigation visualization 多视角无遮挡城市导航可视化

IF 8.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2026-01-07 DOI: 10.1016/j.jag.2025.105074

Shen Ying, Haoran Yang, Junru Su

3D navigation, by virtue of its realism, provides rich, authentic, and visually immersive information that closely approximates the physical world, playing a crucial role in domains such as commerce and scientific research. However, current 3D navigation systems face challenges including the separation of multi-view information and the occlusion of navigation routes. This paper presents an occlusion-free, multi-perspective visualization method for urban navigation routes. We introduce an undulating curve surface and, through computation of an information detail metric, demonstrate that our surface enables rapid transitions from the focus to the context and supports visualization of much larger scene extents within the same screen space, compared to a monotonic cylindrical surface. We further enhance scene information delivery by segmenting the navigation route and adjusting the camera’s tilt angle and field of view to match different surface parameters. To eliminate occlusion of the navigation route, we adjust the position and scale of obstructing buildings, assigning distance-based weights. Compared to the fixed-weight approach, our distance-weighted approach highlights building position and shape features more effectively across varied scenarios. The proposed method is validated using an experimental scenario in a region of Barcelona. Experimental results indicate that our approach can effectively represent scenes at different scales while prominently displaying the navigation route, thereby enhancing the overall navigation experience.

3D导航凭借其真实感，提供了丰富、真实和视觉上身临其境的信息，与现实世界非常接近，在商业和科学研究等领域发挥着至关重要的作用。然而，当前的三维导航系统面临着多视图信息分离和导航路线遮挡等问题。提出了一种无遮挡、多视角的城市导航路线可视化方法。我们引入了一个起伏曲线表面，并通过计算信息细节度量，证明了我们的表面能够从焦点快速过渡到上下文，并支持在相同屏幕空间内可视化更大的场景范围，而不是单调的圆柱形表面。我们通过分割导航路线和调整相机的倾斜角度和视场来匹配不同的表面参数，进一步增强场景信息的传递。为了消除对导航路线的遮挡，我们调整了阻碍建筑物的位置和规模，分配了基于距离的权重。与固定权重方法相比，我们的距离加权方法在不同场景下更有效地突出了建筑位置和形状特征。在巴塞罗那地区的一个实验场景中验证了所提出的方法。实验结果表明，我们的方法可以有效地呈现不同尺度的场景，同时突出显示导航路线，从而增强整体导航体验。

{"title":"Multi-perspective occlusion-free urban navigation visualization","authors":"Shen Ying, Haoran Yang, Junru Su","doi":"10.1016/j.jag.2025.105074","DOIUrl":"10.1016/j.jag.2025.105074","url":null,"abstract":"<div><div>3D navigation, by virtue of its realism, provides rich, authentic, and visually immersive information that closely approximates the physical world, playing a crucial role in domains such as commerce and scientific research. However, current 3D navigation systems face challenges including the separation of multi-view information and the occlusion of navigation routes. This paper presents an occlusion-free, multi-perspective visualization method for urban navigation routes. We introduce an undulating curve surface and, through computation of an information detail metric, demonstrate that our surface enables rapid transitions from the focus to the context and supports visualization of much larger scene extents within the same screen space, compared to a monotonic cylindrical surface. We further enhance scene information delivery by segmenting the navigation route and adjusting the camera’s tilt angle and field of view to match different surface parameters. To eliminate occlusion of the navigation route, we adjust the position and scale of obstructing buildings, assigning distance-based weights. Compared to the fixed-weight approach, our distance-weighted approach highlights building position and shape features more effectively across varied scenarios. The proposed method is validated using an experimental scenario in a region of Barcelona. Experimental results indicate that our approach can effectively represent scenes at different scales while prominently displaying the navigation route, thereby enhancing the overall navigation experience.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"146 ","pages":"Article 105074"},"PeriodicalIF":8.6,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145926328","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}

引用次数: 0