Remote Sensing of Environment最新文献_第7页

The impact of leaf-wood separation algorithms on aboveground biomass estimation from terrestrial laser scanning 叶木分离算法对地面激光扫描估算地上生物量的影响

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-25 DOI: 10.1016/j.rse.2024.114581

Shilin Chen , Hans Verbeeck , Louise Terryn , Wouter A.J. Van den Broeck , Matheus Boni Vicari , Mathias Disney , Niall Origo , Di Wang , Zhouxin Xi , Chris Hopkinson , Wenxia Dai , Meilian Wang , Sruthi M. Krishna Moorthy , Jie Shao , Roberto Ferrara , David W. MacFarlane , Kim Calders

Leaf-wood separation plays an important role in estimating aboveground biomass (AGB) of trees from terrestrial laser scanning (TLS) point clouds. Yet, leaf-wood separation studies have predominantly focused on reporting the accuracy of leaf and wood point separation. Assessments of the impact of these algorithms on the subsequent AGB estimations, based on commonly used quantitative structure models (QSMs), have been limited. Therefore, in this study, we quantified the impact of 11 published leaf-wood separation algorithms on QSM-based tree AGB estimation using an independent benchmarking dataset. The benchmarking dataset consists of AGB measured for 20 destructively harvested trees from a mixed temperate forest in Harvard Forest and AGB estimated from QSMs built on manually segmented tree point clouds of 856 broadleaved trees in Wytham Woods under leaf-off conditions. These benchmarking AGB values were compared to the AGB estimated from QSMs built on the leaf-removed point clouds resulting from the different separation algorithms performed on the leaf-on tree point clouds of the same trees. The results of the study indicated that for most of the algorithms, the leaf-removed AGB estimates for both coniferous and broadleaved trees underestimated the AGB (conifers: −17 % to −3 %, broadleaf: −14 % to −2 %) compared to the destructively measured AGB in Harvard Forest. In Wytham Woods, leaf-removed AGB estimates from all separation algorithms consistently underestimated the AGB (−46 % to −24 %) compared to the AGB from the leaf-off point clouds. Most leaf-wood separation algorithms performed better on broadleaved trees than on coniferous trees. Moreover, significant differences were observed among different algorithms in estimating AGB for trees of the same forest type. For coniferous trees, the relative difference (RD) of leaf-removed AGB estimates from QSMs and separation algorithms ranged from −27 % to 16 %, among which the best performing algorithms demonstrated similar optimal performance, with small RD values of approximately −3 % to 2 %. For broadleaved trees, the leaf-removed AGB estimates from QSMs and eight separation algorithms, as well as leaf-off point cloud estimates (approximately at 10 %), were closely in agreement with the harvested benchmark values, among which the best performing algorithms had a RD value approximately within ±2 %. Additionally, most separation algorithms could lead to better estimates of trunk biomass than branch biomass, whereas the estimation for branch biomass consistently exhibited varying degrees of underestimation. These findings provide a timely reference for utilizing leaf-wood separation algorithms for QSM-based AGB estimation.

叶木分离在利用地面激光扫描（TLS）点云估算树木地上生物量（AGB）中起着重要作用。然而，叶-木分离的研究主要集中在报道叶-木点分离的准确性。基于常用的定量结构模型（qsm），对这些算法对后续AGB估计影响的评估是有限的。因此，在本研究中，我们使用独立的基准数据集量化了11种已发表的叶木分离算法对基于qsm的树木AGB估计的影响。基准数据集由哈佛森林混合温带森林中20棵破坏性采伐树木的AGB测量值和Wytham森林中856棵阔叶树在落叶条件下的人工分割树点云的QSMs估计的AGB组成。将这些基准AGB值与在同一棵树的叶子上的树点云上执行不同分离算法所产生的叶片上的点云上建立的QSMs估计的AGB进行比较。研究结果表明，对于大多数算法，与哈佛森林中破坏性测量的AGB相比，对针叶树和阔叶树的叶片去除AGB估计低估了AGB（针叶树：- 17%至- 3%，阔叶树：- 14%至- 2%）。在威瑟姆森林中，与落叶点云的AGB相比，所有分离算法中去除叶片的AGB估计始终低估了AGB（- 46%至- 24%）。大多数叶-木分离算法在阔叶树上比在针叶树上表现更好。此外，不同算法对同一森林类型树木的AGB估计存在显著差异。对于针叶树，QSMs和分离算法对去叶AGB估计的相对差值（RD）在−27% ~ 16%之间，其中表现最好的算法表现出相似的最优性能，RD值较小，约为−3% ~ 2%。对于阔叶树，QSMs和8种分离算法的去叶AGB估计值以及落叶点云估计值（约为10%）与采伐基准值非常接近，其中表现最好的算法的RD值约在±2%以内。此外，大多数分离算法对树干生物量的估计优于对树枝生物量的估计，而对树枝生物量的估计一直表现出不同程度的低估。这些发现为利用叶木分离算法进行基于qsm的AGB估计提供了及时的参考。

{"title":"The impact of leaf-wood separation algorithms on aboveground biomass estimation from terrestrial laser scanning","authors":"Shilin Chen , Hans Verbeeck , Louise Terryn , Wouter A.J. Van den Broeck , Matheus Boni Vicari , Mathias Disney , Niall Origo , Di Wang , Zhouxin Xi , Chris Hopkinson , Wenxia Dai , Meilian Wang , Sruthi M. Krishna Moorthy , Jie Shao , Roberto Ferrara , David W. MacFarlane , Kim Calders","doi":"10.1016/j.rse.2024.114581","DOIUrl":"10.1016/j.rse.2024.114581","url":null,"abstract":"<div><div>Leaf-wood separation plays an important role in estimating aboveground biomass (AGB) of trees from terrestrial laser scanning (TLS) point clouds. Yet, leaf-wood separation studies have predominantly focused on reporting the accuracy of leaf and wood point separation. Assessments of the impact of these algorithms on the subsequent AGB estimations, based on commonly used quantitative structure models (QSMs), have been limited. Therefore, in this study, we quantified the impact of 11 published leaf-wood separation algorithms on QSM-based tree AGB estimation using an independent benchmarking dataset. The benchmarking dataset consists of AGB measured for 20 destructively harvested trees from a mixed temperate forest in Harvard Forest and AGB estimated from QSMs built on manually segmented tree point clouds of 856 broadleaved trees in Wytham Woods under leaf-off conditions. These benchmarking AGB values were compared to the AGB estimated from QSMs built on the leaf-removed point clouds resulting from the different separation algorithms performed on the leaf-on tree point clouds of the same trees. The results of the study indicated that for most of the algorithms, the leaf-removed AGB estimates for both coniferous and broadleaved trees underestimated the AGB (conifers: −17 % to −3 %, broadleaf: −14 % to −2 %) compared to the destructively measured AGB in Harvard Forest. In Wytham Woods, leaf-removed AGB estimates from all separation algorithms consistently underestimated the AGB (−46 % to −24 %) compared to the AGB from the leaf-off point clouds. Most leaf-wood separation algorithms performed better on broadleaved trees than on coniferous trees. Moreover, significant differences were observed among different algorithms in estimating AGB for trees of the same forest type. For coniferous trees, the relative difference (RD) of leaf-removed AGB estimates from QSMs and separation algorithms ranged from −27 % to 16 %, among which the best performing algorithms demonstrated similar optimal performance, with small RD values of approximately −3 % to 2 %. For broadleaved trees, the leaf-removed AGB estimates from QSMs and eight separation algorithms, as well as leaf-off point cloud estimates (approximately at 10 %), were closely in agreement with the harvested benchmark values, among which the best performing algorithms had a RD value approximately within ±2 %. Additionally, most separation algorithms could lead to better estimates of trunk biomass than branch biomass, whereas the estimation for branch biomass consistently exhibited varying degrees of underestimation. These findings provide a timely reference for utilizing leaf-wood separation algorithms for QSM-based AGB estimation.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114581"},"PeriodicalIF":11.1,"publicationDate":"2024-12-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142886719","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Integrating remote sensing with OpenStreetMap data for comprehensive scene understanding through multi-modal self-supervised learning 结合遥感与OpenStreetMap数据，通过多模态自监督学习实现全面的场景理解

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-23 DOI: 10.1016/j.rse.2024.114573

Lubin Bai , Xiuyuan Zhang , Haoyu Wang , Shihong Du

OpenStreetMap (OSM) contains valuable geographic knowledge for remote sensing (RS) interpretation. They can provide correlated and complementary descriptions of a given region. Integrating RS images with OSM data can lead to a more comprehensive understanding of a geographic scene. But due to the significant differences between them, little progress has been made in data fusion for RS and OSM data, and how to extract, interact, and collaborate the information from multiple geographic data sources remains largely unexplored. In this work, we focus on designing a multi-modal self-supervised learning (SSL) approach to fuse RS images and OSM data, which can extract meaningful features from the two complementary data sources in an unsupervised manner, resulting in comprehensive scene understanding. We harmonize the parts of information extraction, interaction, and collaboration for RS and OSM data into a unified SSL framework, named Rose. For information extraction, we start from the complementarity between the two modalities, designing an OSM encoder to harmoniously align with the ViT image encoder. For information interaction, we leverage the spatial correlation between RS and OSM data to guide the cross-attention module, thereby enhancing the information transfer. For information collaboration, we design the joint mask-reconstruction learning strategy to achieve cooperation between the two modalities, which reconstructs the original inputs by referring to information from both sources. The three parts are interlinked and blending seamlessly into a unified framework. Finally, Rose can generate three kinds of representations, i.e., RS feature, OSM feature, and RS-OSM fusion feature, which can be used for multiple downstream tasks. Extensive experiments on land use semantic segmentation, population estimation, and carbon emission estimation tasks demonstrate the multitasking capability, label efficiency, and robustness to noise of Rose. Rose can associate RS images and OSM data at a fine level of granularity, enhancing its effectiveness on fine-grained tasks like land use semantic segmentation. The code can be found at https://github.com/bailubin/Rose.

OpenStreetMap （OSM）包含遥感（RS）解译的宝贵地理知识。它们可以提供给定区域的相关和互补描述。将RS图像与OSM数据相结合可以更全面地了解地理场景。但由于两者之间的显著差异，RS和OSM数据的数据融合进展甚微，如何从多个地理数据源中提取、交互和协作信息仍然是一个很大的未知领域。在这项工作中，我们专注于设计一种多模态自监督学习（SSL）方法来融合RS图像和OSM数据，该方法可以以无监督的方式从两个互补的数据源中提取有意义的特征，从而实现全面的场景理解。我们将RS和OSM数据的信息提取、交互和协作部分协调到一个名为Rose的统一SSL框架中。在信息提取方面，我们从两种模式的互补性出发，设计了一种OSM编码器与ViT图像编码器和谐对齐。在信息交互方面，我们利用RS和OSM数据之间的空间相关性来引导交叉注意模块，从而增强信息传递。在信息协作方面，我们设计了联合掩模重建学习策略，实现了两种模式之间的合作，该策略通过参考两个来源的信息来重建原始输入。这三个部分相互联系，无缝地融合成一个统一的框架。最后，Rose可以生成三种表示，即RS特征、OSM特征和RS-OSM融合特征，可以用于多个下游任务。在土地利用语义分割、人口估计和碳排放估计任务上的大量实验证明了Rose的多任务处理能力、标签效率和对噪声的鲁棒性。Rose可以将RS图像和OSM数据在细粒度级别上关联起来，从而增强其在细粒度任务（如土地使用语义分割）上的有效性。代码可以在https://github.com/bailubin/Rose上找到。

{"title":"Integrating remote sensing with OpenStreetMap data for comprehensive scene understanding through multi-modal self-supervised learning","authors":"Lubin Bai , Xiuyuan Zhang , Haoyu Wang , Shihong Du","doi":"10.1016/j.rse.2024.114573","DOIUrl":"10.1016/j.rse.2024.114573","url":null,"abstract":"<div><div>OpenStreetMap (OSM) contains valuable geographic knowledge for remote sensing (RS) interpretation. They can provide correlated and complementary descriptions of a given region. Integrating RS images with OSM data can lead to a more comprehensive understanding of a geographic scene. But due to the significant differences between them, little progress has been made in data fusion for RS and OSM data, and how to extract, interact, and collaborate the information from multiple geographic data sources remains largely unexplored. In this work, we focus on designing a multi-modal self-supervised learning (SSL) approach to fuse RS images and OSM data, which can extract meaningful features from the two complementary data sources in an unsupervised manner, resulting in comprehensive scene understanding. We harmonize the parts of information extraction, interaction, and collaboration for RS and OSM data into a unified SSL framework, named Rose. For information extraction, we start from the complementarity between the two modalities, designing an OSM encoder to harmoniously align with the ViT image encoder. For information interaction, we leverage the spatial correlation between RS and OSM data to guide the cross-attention module, thereby enhancing the information transfer. For information collaboration, we design the joint mask-reconstruction learning strategy to achieve cooperation between the two modalities, which reconstructs the original inputs by referring to information from both sources. The three parts are interlinked and blending seamlessly into a unified framework. Finally, Rose can generate three kinds of representations, i.e., RS feature, OSM feature, and RS-OSM fusion feature, which can be used for multiple downstream tasks. Extensive experiments on land use semantic segmentation, population estimation, and carbon emission estimation tasks demonstrate the multitasking capability, label efficiency, and robustness to noise of Rose. Rose can associate RS images and OSM data at a fine level of granularity, enhancing its effectiveness on fine-grained tasks like land use semantic segmentation. The code can be found at <span><span>https://github.com/bailubin/Rose</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114573"},"PeriodicalIF":11.1,"publicationDate":"2024-12-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142874651","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Ground surface displacement measurement from SAR imagery using deep learning 利用深度学习从SAR图像中测量地表位移

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-21 DOI: 10.1016/j.rse.2024.114577

Jinwoo Kim , Hyung-Sup Jung , Zhong Lu

Offset tracking using synthetic aperture radar (SAR) amplitude imagery is a valuable technique for detecting large ground displacements. However, the traditional offset tracking methods with the SAR datasets are computationally intensive and require significant time for processing. We have developed a novel cross-connection Siamese ResNet (CC-ResSiamNet). The model leverages multi-kernel offset tracking for preprocessing, followed by deep learning architectures that incorporate U-Net, cross-connections, and residual and attention blocks to predict pixel offsets between two SAR amplitude images. It is trained and tested on 200 K pairs of reference and secondary SAR amplitude images, alongside corresponding target offset data from Alaska's glaciers. The comparative analysis with multiple deep learning models confirmed that our designed model is highly generalizable, achieving rapid convergence, minimal overfitting, and high prediction accuracy. Through multi-scenario inference with glacier movements, earthquakes, and volcanic eruptions worldwide, the model demonstrates strong performance, closely matching the accuracy of traditional methods while offering significantly faster processing times through parallel computing. The model's rapid displacement mapping capability shows particular promise for improving disaster response and near real-time surface monitoring. While the approach encounters challenges in accurately capturing small-scale displacements, it opens new possibilities for SAR-based surface displacement prediction using machine learning. This research highlights the advantages of combining deep learning with SAR imagery for advancing geophysical analysis, with future applications anticipated as more commercial and scientific SAR missions launch globally.

利用合成孔径雷达（SAR）幅值图像进行偏移跟踪是一种有价值的地面大位移检测技术。然而，传统的基于SAR数据集的偏移量跟踪方法计算量大，处理时间长。我们开发了一种新型的交叉连接暹罗ResNet （CC-ResSiamNet）。该模型利用多核偏移跟踪进行预处理，然后采用深度学习架构，结合U-Net、交叉连接、残差和注意块来预测两个SAR振幅图像之间的像素偏移。它在200 K对参考和次级SAR振幅图像以及阿拉斯加冰川相应的目标偏移数据上进行了训练和测试。与多个深度学习模型的对比分析证实了我们设计的模型具有高度的泛化性，实现了快速收敛、最小过拟合和高预测精度。通过对全球冰川运动、地震和火山爆发的多场景推断，该模型显示出强大的性能，与传统方法的精度非常接近，同时通过并行计算提供了显着更快的处理时间。该模型的快速位移映射能力在改善灾害响应和近实时地面监测方面显示出特别的希望。虽然该方法在准确捕获小尺度位移方面遇到了挑战，但它为使用机器学习进行基于sar的地表位移预测开辟了新的可能性。这项研究强调了将深度学习与SAR图像相结合以推进地球物理分析的优势，随着更多商业和科学SAR任务在全球范围内的推出，预计未来的应用将会更加广泛。

{"title":"Ground surface displacement measurement from SAR imagery using deep learning","authors":"Jinwoo Kim , Hyung-Sup Jung , Zhong Lu","doi":"10.1016/j.rse.2024.114577","DOIUrl":"10.1016/j.rse.2024.114577","url":null,"abstract":"<div><div>Offset tracking using synthetic aperture radar (SAR) amplitude imagery is a valuable technique for detecting large ground displacements. However, the traditional offset tracking methods with the SAR datasets are computationally intensive and require significant time for processing. We have developed a novel cross-connection Siamese ResNet (CC-ResSiamNet). The model leverages multi-kernel offset tracking for preprocessing, followed by deep learning architectures that incorporate U-Net, cross-connections, and residual and attention blocks to predict pixel offsets between two SAR amplitude images. It is trained and tested on 200 K pairs of reference and secondary SAR amplitude images, alongside corresponding target offset data from Alaska's glaciers. The comparative analysis with multiple deep learning models confirmed that our designed model is highly generalizable, achieving rapid convergence, minimal overfitting, and high prediction accuracy. Through multi-scenario inference with glacier movements, earthquakes, and volcanic eruptions worldwide, the model demonstrates strong performance, closely matching the accuracy of traditional methods while offering significantly faster processing times through parallel computing. The model's rapid displacement mapping capability shows particular promise for improving disaster response and near real-time surface monitoring. While the approach encounters challenges in accurately capturing small-scale displacements, it opens new possibilities for SAR-based surface displacement prediction using machine learning. This research highlights the advantages of combining deep learning with SAR imagery for advancing geophysical analysis, with future applications anticipated as more commercial and scientific SAR missions launch globally.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114577"},"PeriodicalIF":11.1,"publicationDate":"2024-12-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142867408","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Coupled hydrologic-electromagnetic framework to model permafrost active layer organic soil dielectric properties 耦合水文-电磁框架模拟多年冻土活动层有机土壤介电特性

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-20 DOI: 10.1016/j.rse.2024.114560

Kazem Bakian-Dogaheh , Yuhuan Zhao , John S. Kimball , Mahta Moghaddam

Arctic permafrost soils contain a vast reservoir of soil organic carbon (SOC) vulnerable to increasing mobilization and decomposition from polar warming and permafrost thaw. How these SOC stocks are responding to global warming is uncertain, partly due to a lack of information on the distribution and status of SOC over vast Arctic landscapes. Soil moisture and organic matter vary substantially over the short vertical distance of the permafrost active layer. The hydrological properties of this seasonally thawed soil layer provide insights for understanding the dielectric behavior of water inside the soil matrix, which is key for developing more effective physics-based radar remote sensing retrieval algorithms for large-scale mapping of SOC. This study provides a coupled hydrologic-electromagnetic framework to model the frequency-dependent dielectric behavior of active layer organic soil. For the first time, we present joint measurement and modeling of the water matric potential, dielectric permittivity, and basic physical properties of 66 soil samples collected across the Alaskan Arctic tundra. The matric potential measurement allows for estimating the soil water retention curve, which helps determine the relaxation time through the Eyring equation. The estimated relaxation time of water molecules in soil is then used in the Debye model to predict the water dielectric behavior in soil. A multi-phase dielectric mixing model is applied to incorporate the contribution of various soil components. The resulting organic soil dielectric model accepts saturation water fraction, organic matter content, mineral texture, temperature, and microwave frequency as inputs to calculate the effective soil dielectric characteristic. The developed dielectric model was validated against lab-measured dielectric data for all soil samples and exhibited robust accuracy. We further validated the dielectric model against field-measured dielectric profiles acquired from five sites on the Alaskan North Slope. Model behavior was also compared against other existing dielectric models, and an in-depth discussion on their validity and limitations in permafrost soils is given. The resulting organic soil dielectric model was then integrated with a multi-layer electromagnetic scattering forward model to simulate radar backscatter under a range of soil profile conditions and model parameters. The results indicate that low frequency (P-, L-band) polarimetric synthetic aperture radars (SARs) have the potential to map water and carbon characteristics in permafrost active layer soils using physics-based radar retrieval algorithms.

北极永久冻土土壤含有巨大的土壤有机碳（SOC）库，极易受到极地变暖和永久冻土融化的增加动员和分解的影响。这些有机碳储量如何对全球变暖做出反应尚不确定，部分原因是缺乏有关有机碳在广大北极地区分布和状态的信息。土壤水分和有机质在永久冻土层活动层的短垂直距离上变化很大。这一季节性解冻土层的水文特性为理解土壤基质内水分的介电行为提供了见解，这是开发更有效的基于物理的雷达遥感检索算法用于大规模土壤有机碳制图的关键。本研究提供了一个耦合的水文-电磁框架来模拟活性层有机土壤的频率相关介电行为。本文首次对阿拉斯加北极冻土带66个土壤样品的水基质电位、介电常数和基本物理性质进行了联合测量和建模。基质电位测量可以估计土壤保水曲线，通过Eyring方程确定松弛时间。然后将土壤中水分子的弛豫时间用在德拜模型中来预测土壤中水的介电行为。采用多相介质混合模型，考虑了土壤各组分的贡献。所得的有机土壤介电模型接受饱和含水率、有机质含量、矿物质地、温度和微波频率作为输入，计算土壤有效介电特性。开发的介电模型与实验室测量的所有土壤样品的介电数据进行了验证，并显示出强大的准确性。我们进一步根据在阿拉斯加北坡的五个地点获得的现场测量的介电剖面验证了介电模型。模型的性能也与其他现有的介电模型进行了比较，并深入讨论了它们在多年冻土中的有效性和局限性。将得到的有机土壤介电模型与多层电磁散射正演模型相结合，模拟了一系列土壤剖面条件和模型参数下的雷达后向散射。结果表明，低频（P、l波段）极化合成孔径雷达（sar）具有利用基于物理的雷达检索算法绘制多年冻土活土层土壤水分和碳特征的潜力。

{"title":"Coupled hydrologic-electromagnetic framework to model permafrost active layer organic soil dielectric properties","authors":"Kazem Bakian-Dogaheh , Yuhuan Zhao , John S. Kimball , Mahta Moghaddam","doi":"10.1016/j.rse.2024.114560","DOIUrl":"10.1016/j.rse.2024.114560","url":null,"abstract":"<div><div>Arctic permafrost soils contain a vast reservoir of soil organic carbon (SOC) vulnerable to increasing mobilization and decomposition from polar warming and permafrost thaw. How these SOC stocks are responding to global warming is uncertain, partly due to a lack of information on the distribution and status of SOC over vast Arctic landscapes. Soil moisture and organic matter vary substantially over the short vertical distance of the permafrost active layer. The hydrological properties of this seasonally thawed soil layer provide insights for understanding the dielectric behavior of water inside the soil matrix, which is key for developing more effective physics-based radar remote sensing retrieval algorithms for large-scale mapping of SOC. This study provides a coupled hydrologic-electromagnetic framework to model the frequency-dependent dielectric behavior of active layer organic soil. For the first time, we present joint measurement and modeling of the water matric potential, dielectric permittivity, and basic physical properties of 66 soil samples collected across the Alaskan Arctic tundra. The matric potential measurement allows for estimating the soil water retention curve, which helps determine the relaxation time through the Eyring equation. The estimated relaxation time of water molecules in soil is then used in the Debye model to predict the water dielectric behavior in soil. A multi-phase dielectric mixing model is applied to incorporate the contribution of various soil components. The resulting organic soil dielectric model accepts saturation water fraction, organic matter content, mineral texture, temperature, and microwave frequency as inputs to calculate the effective soil dielectric characteristic. The developed dielectric model was validated against lab-measured dielectric data for all soil samples and exhibited robust accuracy. We further validated the dielectric model against field-measured dielectric profiles acquired from five sites on the Alaskan North Slope. Model behavior was also compared against other existing dielectric models, and an in-depth discussion on their validity and limitations in permafrost soils is given. The resulting organic soil dielectric model was then integrated with a multi-layer electromagnetic scattering forward model to simulate radar backscatter under a range of soil profile conditions and model parameters. The results indicate that low frequency (P-, L-band) polarimetric synthetic aperture radars (SARs) have the potential to map water and carbon characteristics in permafrost active layer soils using physics-based radar retrieval algorithms.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114560"},"PeriodicalIF":11.1,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142857725","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Joint mapping of melt pond bathymetry and water volume on sea ice using optical remote sensing images and physical reflectance models 基于光学遥感影像和物理反射模型的融池水深和海冰水量联合制图

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-20 DOI: 10.1016/j.rse.2024.114571

Chuan Xiong, Xudong Li

Melt ponds are a common phenomenon on the surface of Arctic sea ice during the summer, and their low albedo strongly influences the energy balance of the Arctic sea ice. Estimating Melt Pond Fraction (MPF) and Melt Pond Depth (MPD) using optical remote sensing is crucial for a better understanding of rapid climate change in the Arctic region. However, current retrieval algorithms for monitoring Arctic melt ponds using optical imagery often fail to estimate MPD. In this study, a radiative transfer model for melt ponds is establish to describe the relationship between melt pond reflectance and its physical properties. Using Sentinel-2 observation data, we propose a novel algorithm for the simultaneous retrieval of MPF and MPD, thereby enabling the estimation of Melt Pond Volume (MPV). This method does not depend on prior assumptions regarding the spectral reflectance of sea ice and melt ponds, and it accounts for the spatiotemporal variability in their reflectance. Compared with other high-resolution MPF and MPD products, the results of this study demonstrate comparable spatial distributions. The root mean square error (RMSE) of the retrieved MPF is less than 10 %, and the RMSE for MPD is approximately 24.51 cm. The analysis of melt pond evolution along the MOSAiC track shows the rapid expansion of melt ponds and their significant spatial variability. Ultimately, using Google Earth Engine (GEE) and machine learning, a dataset of MPF, MPD, and MPV for the Arctic from 2013 to 2023 is generated from 57,842 Landsat-8 images. Correlation analysis shows that MPF, MPD, and MPV all have a positive correlation with downward surface radiation. The approach outlined in this study is entirely based on remote sensing imagery, demonstrating significant potential for large scale application. This offers new opportunities for estimating the volume of water stored in Arctic summer melt ponds.

融池是夏季北极海冰表面普遍存在的现象，其低反照率强烈影响着北极海冰的能量平衡。利用光学遥感估算融池分数（MPF）和融池深度（MPD）对于更好地了解北极地区的快速气候变化至关重要。然而，目前使用光学图像监测北极融化池的检索算法往往无法估计MPD。本文建立了熔池辐射传输模型，描述了熔池反射率与其物理性质之间的关系。利用Sentinel-2观测数据，提出了一种同时检索MPF和MPD的新算法，从而实现了融池体积（MPV）的估算。该方法不依赖于对海冰和融化池光谱反射率的先验假设，并且考虑了其反射率的时空变异性。与其他高分辨率MPF和MPD产品相比，本研究结果具有可比性。反演MPF的均方根误差（RMSE）小于10%，MPD的RMSE约为24.51 cm。在MOSAiC轨迹上对熔池演化的分析表明，熔池扩张迅速，空间变异性显著。最终，利用谷歌地球引擎（GEE）和机器学习，从57,842张Landsat-8图像中生成了2013年至2023年北极的MPF、MPD和MPV数据集。相关分析表明，MPF、MPD和MPV均与地表向下辐射呈正相关。本研究概述的方法完全基于遥感图像，显示出大规模应用的巨大潜力。这为估计北极夏季融化池中储存的水量提供了新的机会。

{"title":"Joint mapping of melt pond bathymetry and water volume on sea ice using optical remote sensing images and physical reflectance models","authors":"Chuan Xiong, Xudong Li","doi":"10.1016/j.rse.2024.114571","DOIUrl":"10.1016/j.rse.2024.114571","url":null,"abstract":"<div><div>Melt ponds are a common phenomenon on the surface of Arctic sea ice during the summer, and their low albedo strongly influences the energy balance of the Arctic sea ice. Estimating Melt Pond Fraction (MPF) and Melt Pond Depth (MPD) using optical remote sensing is crucial for a better understanding of rapid climate change in the Arctic region. However, current retrieval algorithms for monitoring Arctic melt ponds using optical imagery often fail to estimate MPD. In this study, a radiative transfer model for melt ponds is establish to describe the relationship between melt pond reflectance and its physical properties. Using Sentinel-2 observation data, we propose a novel algorithm for the simultaneous retrieval of MPF and MPD, thereby enabling the estimation of Melt Pond Volume (MPV). This method does not depend on prior assumptions regarding the spectral reflectance of sea ice and melt ponds, and it accounts for the spatiotemporal variability in their reflectance. Compared with other high-resolution MPF and MPD products, the results of this study demonstrate comparable spatial distributions. The root mean square error (RMSE) of the retrieved MPF is less than 10 %, and the RMSE for MPD is approximately 24.51 cm. The analysis of melt pond evolution along the MOSAiC track shows the rapid expansion of melt ponds and their significant spatial variability. Ultimately, using Google Earth Engine (GEE) and machine learning, a dataset of MPF, MPD, and MPV for the Arctic from 2013 to 2023 is generated from 57,842 Landsat-8 images. Correlation analysis shows that MPF, MPD, and MPV all have a positive correlation with downward surface radiation. The approach outlined in this study is entirely based on remote sensing imagery, demonstrating significant potential for large scale application. This offers new opportunities for estimating the volume of water stored in Arctic summer melt ponds.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114571"},"PeriodicalIF":11.1,"publicationDate":"2024-12-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142857729","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A flexible framework for built-up height mapping using ICESat-2 photons and multisource satellite observations 使用ICESat-2光子和多源卫星观测进行建筑高度制图的灵活框架

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-19 DOI: 10.1016/j.rse.2024.114572

Xiayu Tang , Guojiang Yu , Xuecao Li , Hannes Taubenböck , Guohua Hu , Yuyu Zhou , Cong Peng , Donglie Liu , Jianxi Huang , Xiaoping Liu , Peng Gong

Built-up heights serve as a nexus in understanding the complex relationship between urban forms and socioeconomic activities. With the advent of remote sensing technology, built-up height mapping from satellite observations has become available over the past years. However, the absence of high-precision sample data poses a significant limitation to built-up height mapping at large (regional or global) scales, particularly in developing regions. To address this issue, we proposed a flexible mapping framework to derive precise building height samples using the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) data for built-up height estimation. First, we calculated building heights from ICESat-2 photons using advanced algorithms such as Random Sample Consensus (RANSAC) linear fitting and cloth simulation filtering. Then, we constructed large-scale built-up height samples by aggregating the height information into grid cells with optimal size. Finally, aided by these grids with height information from ICEsat-2 and other satellite observations from Sentinel data as well as the digital surface model (DSM), we mapped built-up heights in two mega-cities (i.e., New York and Shenzhen) using the random forest regression model. Our results demonstrate building height estimation using ICESat-2 data generally exhibits in relation to other studies high accuracy, showing great potential to support large-scale built-up height mapping using satellite observations. We found the optimal grid size for built-up height mapping is around 300 m, after a comprehensive sensitivity analysis regarding the building fraction within the grid and its size. Overall, the mapped built-up heights are reliable, with relatively low mean absolute errors (MAE) of 2.69 m in New York and 3.87 m in Shenzhen, similar to or better than previous studies. By leveraging high-precision elevation data provided by the ICESat-2 data, our proposed approach can effectively collect samples in regions with limited information on building heights, showing great potential for large-scale built-up height monitoring and supporting future urban studies.

建筑高度是理解城市形态和社会经济活动之间复杂关系的纽带。随着遥感技术的出现，从卫星观测得到的建筑高度图在过去几年已经成为可能。然而，缺乏高精度样本数据对大尺度（区域或全球）建筑高度制图造成了重大限制，特别是在发展中地区。为了解决这个问题，我们提出了一个灵活的制图框架，利用冰、云和陆地高程卫星-2 （ICESat-2）数据获得精确的建筑物高度样本，用于建筑物高度估计。首先，我们使用随机样本共识（RANSAC）线性拟合和布模拟滤波等先进算法从ICESat-2光子计算建筑物高度。然后，将高度信息聚合到最优大小的网格单元中，构建大规模建筑高度样本；最后，借助ICEsat-2和Sentinel数据的其他卫星观测数据以及数字地面模型（DSM）提供的网格高度信息，我们使用随机森林回归模型绘制了两个特大城市（即纽约和深圳）的建筑高度图。我们的研究结果表明，与其他研究相比，使用ICESat-2数据估算建筑物高度通常具有较高的精度，显示出支持使用卫星观测进行大规模建筑物高度制图的巨大潜力。在对网格内的建筑比例及其大小进行综合敏感性分析后，我们发现建筑物高度映射的最佳网格尺寸约为300米。总体而言，地图上的建筑高度是可靠的，纽约和深圳的平均绝对误差（MAE）相对较低，分别为2.69 m和3.87 m，与之前的研究相似或更好。通过利用ICESat-2数据提供的高精度高程数据，我们提出的方法可以有效地在建筑高度信息有限的地区收集样本，显示出大规模建筑高度监测和支持未来城市研究的巨大潜力。

{"title":"A flexible framework for built-up height mapping using ICESat-2 photons and multisource satellite observations","authors":"Xiayu Tang , Guojiang Yu , Xuecao Li , Hannes Taubenböck , Guohua Hu , Yuyu Zhou , Cong Peng , Donglie Liu , Jianxi Huang , Xiaoping Liu , Peng Gong","doi":"10.1016/j.rse.2024.114572","DOIUrl":"10.1016/j.rse.2024.114572","url":null,"abstract":"<div><div>Built-up heights serve as a nexus in understanding the complex relationship between urban forms and socioeconomic activities. With the advent of remote sensing technology, built-up height mapping from satellite observations has become available over the past years. However, the absence of high-precision sample data poses a significant limitation to built-up height mapping at large (regional or global) scales, particularly in developing regions. To address this issue, we proposed a flexible mapping framework to derive precise building height samples using the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) data for built-up height estimation. First, we calculated building heights from ICESat-2 photons using advanced algorithms such as Random Sample Consensus (RANSAC) linear fitting and cloth simulation filtering. Then, we constructed large-scale built-up height samples by aggregating the height information into grid cells with optimal size. Finally, aided by these grids with height information from ICEsat-2 and other satellite observations from Sentinel data as well as the digital surface model (DSM), we mapped built-up heights in two mega-cities (i.e., New York and Shenzhen) using the random forest regression model. Our results demonstrate building height estimation using ICESat-2 data generally exhibits in relation to other studies high accuracy, showing great potential to support large-scale built-up height mapping using satellite observations. We found the optimal grid size for built-up height mapping is around 300 m, after a comprehensive sensitivity analysis regarding the building fraction within the grid and its size. Overall, the mapped built-up heights are reliable, with relatively low mean absolute errors (MAE) of 2.69 m in New York and 3.87 m in Shenzhen, similar to or better than previous studies. By leveraging high-precision elevation data provided by the ICESat-2 data, our proposed approach can effectively collect samples in regions with limited information on building heights, showing great potential for large-scale built-up height monitoring and supporting future urban studies.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114572"},"PeriodicalIF":11.1,"publicationDate":"2024-12-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142857723","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Unveiling multimodal consolidation process of the newly reclaimed HKIA 3rd runway from satellite SAR interferometry, ICA analytics and Terzaghi consolidation theory 利用卫星SAR干涉测量、ICA分析及Terzaghi固结理论，揭示新填海的香港国际机场第三跑道的多模式固结过程

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-17 DOI: 10.1016/j.rse.2024.114561

Zhuo Jiang , Guoqiang Shi , Songbo Wu , Xiaoli Ding , Chaoying Zhao , Man Sing Wong , Zhong Lu

The three-runway system expansion project of the Hong Kong International Airport (HKIA) began with the land reclamation to the north of its original runway. To facilitate quick stabilization, the Deep Cement Mixing (DCM) in this project was featured as the novel reclamation method firstly applied in Hong Kong. Understanding ground deformation and underground consolidation is crucial for subsequent soil improvement, civil construction, and future planning at the new platform. Synthetic Aperture Radar Interferometry (InSAR) is used to investigate the spatiotemporal characteristics of land deformation following the completion of the third runway pavement. A combined strategy of persistent scatterer (PS) and distributed scatterer (DS) interferometry was implemented to address low radar coherence at the site. The new reclamation is subject to varying degrees of land subsidence, with a maximum monitored sinking rate to be ∼150 mm/year during September 2021 and October 2023. Whereas the 3rd runway was urgently transformed to operation, spatial details of consolidation status of this new land were not yet evaluated. We applied the Independent Component Analysis (ICA) to identify the underlying sources leading to the measured deformation from InSAR. Three distinct sources have been unveiled, including an exponential decay signal (a quick compaction subsidence of surficial materials), a linear signal (a continuous subsiding from marine deposits) and a periodic signal (thermal effects correlated with buildings and bridges). Notably, the linear deformation component is mainly located in areas with prefabricated vertical drains (PVD), which is strongly correlating with the current monitored subsidence pattern. We incorporated the Terzaghi consolidation theory to further characterize InSAR displacement and estimate the subsidence decay property, consolidation time, ultimate primary settlement and consolidation degree at the 3rd runway, with unprecedented spatial details. Our results indicate the DCM method achieves geological stability more rapidly than the PVD method, with a time advantage of approximately 0.08–1.39 years. Meanwhile, DCM can effectively control the primary settlement to 29 % - 83 % of the PVD method. This research benefits our understanding of the consolidation process at the 3rd runway and offer reliable and detailed data of underground properties. This facilitates more accurate planning of follow-up reinforcement measures at specific locations if needed, which also serves as a valuable reference for future reclamation practices in Hong Kong, particularly using the DCM method.

香港国际机场三跑道系统扩建工程首先在原有跑道北面进行填海造地。为了快速稳定土质，该项目中的深层水泥搅拌法（DCM）是香港首次采用的新型填海方法。了解地面变形和地下固结对后续的土壤改良、土木工程和新平台的未来规划至关重要。合成孔径雷达干涉测量法（InSAR）用于研究第三跑道铺设完成后土地变形的时空特征。为解决现场雷达相干性低的问题，采用了持续散射体（PS）和分布式散射体（DS）干涉测量相结合的策略。新填海区存在不同程度的地面沉降，在 2021 年 9 月至 2023 年 10 月期间，监测到的最大下沉速度为每年 150 毫米。虽然第三跑道急需改造投入使用，但尚未对这一新土地的空间固结状况进行详细评估。我们采用独立分量分析（ICA）来确定导致 InSAR 测量变形的基本来源。我们发现了三个不同的来源，包括指数衰减信号（表层材料的快速压实下沉）、线性信号（海洋沉积物的持续下沉）和周期信号（与建筑物和桥梁相关的热效应）。值得注意的是，线性形变部分主要位于预制垂直排水沟（PVD）区域，这与当前监测到的沉降模式密切相关。我们结合特尔扎吉固结理论进一步描述了 InSAR 位移的特征，并估算了第三跑道的沉降衰减特性、固结时间、最终一次沉降和固结程度，其空间细节前所未有。结果表明，DCM 方法比 PVD 方法更快实现地质稳定，时间优势约为 0.08-1.39 年。同时，DCM 可以有效地将一次沉降控制在 PVD 方法的 29% - 83%。这项研究有助于我们了解第三跑道的固结过程，并提供可靠、详细的地下属性数据。這有助我們在有需要時更準確地規劃特定地點的後續加固措施，對香港日後的填海工程，特別是採用 DCM 方法，提供寶貴的參考資料。

{"title":"Unveiling multimodal consolidation process of the newly reclaimed HKIA 3rd runway from satellite SAR interferometry, ICA analytics and Terzaghi consolidation theory","authors":"Zhuo Jiang , Guoqiang Shi , Songbo Wu , Xiaoli Ding , Chaoying Zhao , Man Sing Wong , Zhong Lu","doi":"10.1016/j.rse.2024.114561","DOIUrl":"10.1016/j.rse.2024.114561","url":null,"abstract":"<div><div>The three-runway system expansion project of the Hong Kong International Airport (HKIA) began with the land reclamation to the north of its original runway. To facilitate quick stabilization, the Deep Cement Mixing (DCM) in this project was featured as the novel reclamation method firstly applied in Hong Kong. Understanding ground deformation and underground consolidation is crucial for subsequent soil improvement, civil construction, and future planning at the new platform. Synthetic Aperture Radar Interferometry (InSAR) is used to investigate the spatiotemporal characteristics of land deformation following the completion of the third runway pavement. A combined strategy of persistent scatterer (PS) and distributed scatterer (DS) interferometry was implemented to address low radar coherence at the site. The new reclamation is subject to varying degrees of land subsidence, with a maximum monitored sinking rate to be ∼150 mm/year during September 2021 and October 2023. Whereas the 3rd runway was urgently transformed to operation, spatial details of consolidation status of this new land were not yet evaluated. We applied the Independent Component Analysis (ICA) to identify the underlying sources leading to the measured deformation from InSAR. Three distinct sources have been unveiled, including an exponential decay signal (a quick compaction subsidence of surficial materials), a linear signal (a continuous subsiding from marine deposits) and a periodic signal (thermal effects correlated with buildings and bridges). Notably, the linear deformation component is mainly located in areas with prefabricated vertical drains (PVD), which is strongly correlating with the current monitored subsidence pattern. We incorporated the Terzaghi consolidation theory to further characterize InSAR displacement and estimate the subsidence decay property, consolidation time, ultimate primary settlement and consolidation degree at the 3rd runway, with unprecedented spatial details. Our results indicate the DCM method achieves geological stability more rapidly than the PVD method, with a time advantage of approximately 0.08–1.39 years. Meanwhile, DCM can effectively control the primary settlement to 29 % - 83 % of the PVD method. This research benefits our understanding of the consolidation process at the 3rd runway and offer reliable and detailed data of underground properties. This facilitates more accurate planning of follow-up reinforcement measures at specific locations if needed, which also serves as a valuable reference for future reclamation practices in Hong Kong, particularly using the DCM method.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114561"},"PeriodicalIF":11.1,"publicationDate":"2024-12-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142833054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

How high are we? Large-scale building height estimation at 10 m using Sentinel-1 SAR and Sentinel-2 MSI time series 我们有多高？利用Sentinel-1 SAR和Sentinel-2 MSI时间序列估算10 m的大尺度建筑高度

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-16 DOI: 10.1016/j.rse.2024.114556

Ritu Yadav, Andrea Nascetti, Yifang Ban

Accurate building height estimation is essential to support urbanization monitoring, environmental impact analysis and sustainable urban planning. However, conducting large-scale building height estimation remains a significant challenge. While deep learning (DL) has proven effective for large-scale mapping tasks, there is a lack of advanced DL models specifically tailored for height estimation, particularly when using open-source Earth observation data. In this study, we propose T-SwinUNet, an advanced DL model for large-scale building height estimation leveraging Sentinel-1 SAR and Sentinel-2 multispectral time series. T-SwinUNet model contains a feature extractor with local/global feature comprehension capabilities, a temporal attention module to learn the correlation between constant and variable features of building objects over time and an efficient multitask decoder to predict building height at 10 m spatial resolution. The model is trained and evaluated on data from the Netherlands, Switzerland, Estonia, and Germany, and its generalizability is evaluated on an out-of-distribution (OOD) test set from ten additional cities from other European countries. Our study incorporates extensive model evaluations, ablation experiments, and comparisons with established models. T-SwinUNet predicts building height with a Root Mean Square Error (RMSE) of 1.89 m, outperforming state-of-the-art models at 10 m spatial resolution. Its strong generalization to the OOD test set (RMSE of 3.2 m) underscores its potential for low-cost building height estimation across Europe, with future scalability to other regions. Furthermore, the assessment at 100 m resolution reveals that T-SwinUNet (0.29 m RMSE, 0.75

R^{2}

) also outperformed the global building height product GHSL-Built-H R2023A product(0.56 m RMSE and 0.37

R^{2}

). Our implementation is available at: https://github.com/RituYadav92/Building-Height-Estimation.

准确的建筑高度估算对于支持城市化监测、环境影响分析和可持续城市规划至关重要。然而，进行大规模建筑高度估算仍然是一项重大挑战。虽然深度学习（DL）已被证明对大规模测绘任务非常有效，但目前还缺乏专门用于高度估算的高级 DL 模型，尤其是在使用开源地球观测数据时。在本研究中，我们提出了 T-SwinUNet，一种利用 Sentinel-1 SAR 和 Sentinel-2 多光谱时间序列进行大规模建筑物高度估算的高级 DL 模型。T-SwinUNet 模型包含一个具有局部/全局特征理解能力的特征提取器、一个用于学习建筑物体随时间变化的恒定和可变特征之间的相关性的时间关注模块，以及一个用于预测 10 米空间分辨率下建筑高度的高效多任务解码器。该模型在荷兰、瑞士、爱沙尼亚和德国的数据上进行了训练和评估，并在来自其他欧洲国家另外十个城市的分布外（OOD）测试集上对其通用性进行了评估。我们的研究包括广泛的模型评估、消融实验以及与已有模型的比较。T-SwinUNet 预测建筑高度的均方根误差 (RMSE) 为 1.89 米，在 10 米空间分辨率下优于最先进的模型。它对 OOD 测试集（均方根误差为 3.2 米）的强大普适性强调了其在欧洲低成本建筑高度估算方面的潜力，未来还可扩展到其他地区。此外，在 100 米分辨率下进行的评估显示，T-SwinUNet（0.29 米均方根误差，0.75 R2R2）也优于全球建筑高度产品 GHSL-Built-H R2023A 产品（0.56 米均方根误差，0.37 R2R2）。我们的实施方案可在以下网址获取：https://github.com/RituYadav92/Building-Height-Estimation.

{"title":"How high are we? Large-scale building height estimation at 10 m using Sentinel-1 SAR and Sentinel-2 MSI time series","authors":"Ritu Yadav, Andrea Nascetti, Yifang Ban","doi":"10.1016/j.rse.2024.114556","DOIUrl":"10.1016/j.rse.2024.114556","url":null,"abstract":"<div><div>Accurate building height estimation is essential to support urbanization monitoring, environmental impact analysis and sustainable urban planning. However, conducting large-scale building height estimation remains a significant challenge. While deep learning (DL) has proven effective for large-scale mapping tasks, there is a lack of advanced DL models specifically tailored for height estimation, particularly when using open-source Earth observation data. In this study, we propose T-SwinUNet, an advanced DL model for large-scale building height estimation leveraging Sentinel-1 SAR and Sentinel-2 multispectral time series. T-SwinUNet model contains a feature extractor with local/global feature comprehension capabilities, a temporal attention module to learn the correlation between constant and variable features of building objects over time and an efficient multitask decoder to predict building height at 10 m spatial resolution. The model is trained and evaluated on data from the Netherlands, Switzerland, Estonia, and Germany, and its generalizability is evaluated on an out-of-distribution (OOD) test set from ten additional cities from other European countries. Our study incorporates extensive model evaluations, ablation experiments, and comparisons with established models. T-SwinUNet predicts building height with a Root Mean Square Error (RMSE) of 1.89 m, outperforming state-of-the-art models at 10 m spatial resolution. Its strong generalization to the OOD test set (RMSE of 3.2 m) underscores its potential for low-cost building height estimation across Europe, with future scalability to other regions. Furthermore, the assessment at 100 m resolution reveals that T-SwinUNet (0.29 m RMSE, 0.75 <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>) also outperformed the global building height product GHSL-Built-H R2023A product(0.56 m RMSE and 0.37 <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>). Our implementation is available at: <span><span>https://github.com/RituYadav92/Building-Height-Estimation</span><svg><path></path></svg></span>.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114556"},"PeriodicalIF":11.1,"publicationDate":"2024-12-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142825602","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A radiative transfer model for characterizing photometric and polarimetric properties of leaf reflection: Combination of PROSPECT and a polarized reflection function 表征叶片反射光度和偏振特性的辐射传输模型：PROSPECT和偏振反射函数的组合

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-14 DOI: 10.1016/j.rse.2024.114559

Xiao Li , Zhongqiu Sun , Shan Lu , Kenji Omasa

Light photometric and polarimetric characteristics are crucial for describing the optical properties of leaf reflections, which play an essential role in investigating biochemical and surface structural trait inversion and radiative balance between vegetation and atmospheric system. Although several physical models are available, research on a comprehensive model that accounts for both photometric and polarimetric characteristics and incorporates biochemical and surface structural traits is still inadequate. In this study, we introduced PROPOLAR, a leaf model that considered leaf reflection in terms of polarized and unpolarized components and linked leaf reflection to leaf traits. PROPOLAR employed PROSPECT to simulate non-polarized component associated with biochemical traits, while used a three-parameter function (linear coefficient, refractive index factor, and roughness of leaf surface) to simulate the polarized component. The model was validated using a dataset (composed of both photometric and polarimetric measurements) collected from 533 samples of 13 plant species under various illumination-viewing geometries. The results showed that PROPOLAR outperformed PROSPECT and PROSPECULAR (a leaf model charactering BRF) in simulating light intensity (R² = 0.98), and effectively simulated bidirectional polarization reflectance factor (BPRF) and degree of linear polarization (Dolp) across a wide spectral range (450–2300 nm) and species, with R² = 0.92, and 0.80, respectively. Furthermore, PROPOLAR enhanced the accuracy of PROSPECT and showed comparable accuracy with PROSPECULAR in the inversion of biochemical traits from the multi-angular polarization measurements, including chlorophyll (R² = 0.89, RMSE = 12.83 μg/cm²), equivalent water thickness (R² = 0.90, RMSE = 0.0032 g/cm²), and leaf mass per area (R² = 0.38, RMSE = 0.0031 g/cm²), due to the incorporation of polarization reflection and a linear coefficient during calibration. Notably, PROPOLAR can invert roughness and showed reasonable consistency with measured roughness (R² = 0.61). These results demonstrated the effectiveness of PROPOLAR in simulating both photometric and polarimetric properties of leaf reflection, as well as its potential for biochemical and surface structural trait inversion. PROPOLAR may advance remote sensing applications in vegetation management by integrating photometric and polarimetric properties.

光的光度和偏振特性是描述叶片反射光学特性的关键，在研究植被与大气系统的生物化学和表面结构特征反演以及辐射平衡等方面发挥着重要作用。虽然有几种物理模型可用，但考虑到光度和极化特征并结合生化和表面结构特征的综合模型的研究仍然不足。在本研究中，我们引入了PROPOLAR叶片模型，该模型考虑了叶片反射的极化和非极化分量，并将叶片反射与叶片性状联系起来。PROPOLAR采用PROSPECT模拟与生化性状相关的非极化成分，采用线性系数、折射率因子和叶片表面粗糙度三参数函数模拟极化成分。该模型使用从13种植物的533个样品中收集的数据集（由光度和偏振测量组成）在不同的照明观测几何形状下进行验证。结果表明，PROPOLAR在模拟光强方面优于PROSPECT和PROSPECULAR（表征BRF的叶片模型）（R2 = 0.98），在较宽光谱范围（450 ~ 2300 nm）和物种范围内有效模拟双向偏振反射因子（BPRF）和线性偏振度（Dolp）， R2分别为0.92和0.80。此外，由于在校正过程中加入了偏振反射和线性系数，PROPOLAR提高了PROSPECT的精度，在多角度偏振测量的生化性状反演中，包括叶绿素（R2 = 0.89, RMSE = 12.83 μg/cm2）、等效水厚度（R2 = 0.90, RMSE = 0.0032 g/cm2）和每面积叶质量（R2 = 0.38, RMSE = 0.0031 g/cm2），其精度与PROSPECULAR相当。值得注意的是，PROPOLAR可以反演粗糙度，并且与实测粗糙度具有合理的一致性（R2 = 0.61）。这些结果证明了PROPOLAR在模拟叶片反射的光度和偏振特性方面的有效性，以及它在生化和表面结构特征反演方面的潜力。PROPOLAR可以通过整合光度和偏振特性来推进遥感在植被管理中的应用。

{"title":"A radiative transfer model for characterizing photometric and polarimetric properties of leaf reflection: Combination of PROSPECT and a polarized reflection function","authors":"Xiao Li , Zhongqiu Sun , Shan Lu , Kenji Omasa","doi":"10.1016/j.rse.2024.114559","DOIUrl":"10.1016/j.rse.2024.114559","url":null,"abstract":"<div><div>Light photometric and polarimetric characteristics are crucial for describing the optical properties of leaf reflections, which play an essential role in investigating biochemical and surface structural trait inversion and radiative balance between vegetation and atmospheric system. Although several physical models are available, research on a comprehensive model that accounts for both photometric and polarimetric characteristics and incorporates biochemical and surface structural traits is still inadequate. In this study, we introduced PROPOLAR, a leaf model that considered leaf reflection in terms of polarized and unpolarized components and linked leaf reflection to leaf traits. PROPOLAR employed PROSPECT to simulate non-polarized component associated with biochemical traits, while used a three-parameter function (linear coefficient, refractive index factor, and roughness of leaf surface) to simulate the polarized component. The model was validated using a dataset (composed of both photometric and polarimetric measurements) collected from 533 samples of 13 plant species under various illumination-viewing geometries. The results showed that PROPOLAR outperformed PROSPECT and PROSPECULAR (a leaf model charactering BRF) in simulating light intensity (R<sup>2</sup> = 0.98), and effectively simulated bidirectional polarization reflectance factor (BPRF) and degree of linear polarization (Dolp) across a wide spectral range (450–2300 nm) and species, with R<sup>2</sup> = 0.92, and 0.80, respectively. Furthermore, PROPOLAR enhanced the accuracy of PROSPECT and showed comparable accuracy with PROSPECULAR in the inversion of biochemical traits from the multi-angular polarization measurements, including chlorophyll (R<sup>2</sup> = 0.89, RMSE = 12.83 μg/cm<sup>2</sup>), equivalent water thickness (R<sup>2</sup> = 0.90, RMSE = 0.0032 g/cm<sup>2</sup>), and leaf mass per area (R<sup>2</sup> = 0.38, RMSE = 0.0031 g/cm<sup>2</sup>), due to the incorporation of polarization reflection and a linear coefficient during calibration. Notably, PROPOLAR can invert roughness and showed reasonable consistency with measured roughness (R<sup>2</sup> = 0.61). These results demonstrated the effectiveness of PROPOLAR in simulating both photometric and polarimetric properties of leaf reflection, as well as its potential for biochemical and surface structural trait inversion. PROPOLAR may advance remote sensing applications in vegetation management by integrating photometric and polarimetric properties.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114559"},"PeriodicalIF":11.1,"publicationDate":"2024-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142820925","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Predicting drought vulnerability with leaf reflectance spectra in Amazonian trees 利用亚马逊树木的叶片反射光谱预测干旱脆弱性

IF 11.1 1区地球科学 Q1 ENVIRONMENTAL SCIENCES

Remote Sensing of Environment

Pub Date : 2024-12-14 DOI: 10.1016/j.rse.2024.114562

Maquelle N. Garcia , Lucas B.S. Tameirão , Juliana Schietti , Izabela Aleixo , Tomas F. Domingues , K. Fred Huemmrich , Petya K.E. Campell , Loren P. Albert

Hydraulic traits mediate trade-offs between growth and mortality in plants yet characterizing these traits at the community level remains challenging, particularly in the Amazon, where they vary widely across species and environments. While previous studies have used reflectance-based estimates, hydraulic traits, which arise from wood and/or whole-plant anatomy and physiology, have not been comprehensively explored.

For the first time, we comprehensively investigated the use of leaf reflectance to predict hydraulic traits alongside leaf functional traits in tropical evergreen and deciduous trees. For 196 Amazonian trees, we measured water potential, leaf mass per area (LMA), leaf reflectance, hydraulic conductivity curves (e.g., P50), and wood density (WD). We examined the relationships between leaf reflectance and traits using partial least square regression (PLSR).

Our findings indicate that leaf reflectance accurately predicts variation in LMA (R² = 0.8), and reasonably estimates xylem water potential (R² = 0.51) and WD (R² = 0.52). However, P50 predictions were much less reliable (R² = 0.27), with water absorption bands greatly influencing the PLSR model. Leaf phenological strategy had little impact on the results.

These findings suggest that reflectance-based remote sensing could monitor water status and forest carbon dynamics through water potential and wood density, respectively. However, our case study applying the PLSR approach to hyperspectral canopy spectra to predict wood density revealed challenges to upscaling. Despite these limitations, remote sensing of forest hydraulic traits at scale could enhance our understanding of drought vulnerability and carbon dynamics in Amazonian forests, with significant implications for conservation.

水力性状调节植物生长和死亡之间的权衡，但在群落水平上表征这些性状仍然具有挑战性，特别是在亚马逊地区，它们在不同物种和环境中差异很大。虽然以前的研究使用了基于反射率的估计，但由于木材和/或整株植物的解剖和生理，水力特性尚未得到全面探索。本文首次全面研究了利用叶片反射率预测热带常绿和落叶乔木水力性状和叶片功能性状的方法。对于196棵亚马逊树，我们测量了水势、每面积叶质量（LMA）、叶片反射率、水力导率曲线（如P50）和木材密度（WD）。利用偏最小二乘法（PLSR）分析了叶片反射率与性状之间的关系。结果表明，叶片反射率能准确预测叶片LMA的变化（R2 = 0.8），能合理预测木质部水势（R2 = 0.51）和WD （R2 = 0.52）。然而，P50预测的可靠性要低得多（R2 = 0.27），吸水带对PLSR模型的影响很大。叶片物候策略对结果影响不大。这些结果表明，基于反射率的遥感可以分别通过水势和木材密度监测水分状况和森林碳动态。然而，我们的案例研究将PLSR方法应用于高光谱冠层光谱来预测木材密度，揭示了升级的挑战。尽管存在这些限制，大规模的森林水力特征遥感可以增强我们对亚马逊森林干旱脆弱性和碳动态的理解，对保护具有重要意义。

{"title":"Predicting drought vulnerability with leaf reflectance spectra in Amazonian trees","authors":"Maquelle N. Garcia , Lucas B.S. Tameirão , Juliana Schietti , Izabela Aleixo , Tomas F. Domingues , K. Fred Huemmrich , Petya K.E. Campell , Loren P. Albert","doi":"10.1016/j.rse.2024.114562","DOIUrl":"10.1016/j.rse.2024.114562","url":null,"abstract":"<div><div>Hydraulic traits mediate trade-offs between growth and mortality in plants yet characterizing these traits at the community level remains challenging, particularly in the Amazon, where they vary widely across species and environments. While previous studies have used reflectance-based estimates, hydraulic traits, which arise from wood and/or whole-plant anatomy and physiology, have not been comprehensively explored.</div><div>For the first time, we comprehensively investigated the use of leaf reflectance to predict hydraulic traits alongside leaf functional traits in tropical evergreen and deciduous trees. For 196 Amazonian trees, we measured water potential, leaf mass per area (LMA), leaf reflectance, hydraulic conductivity curves (e.g., P50), and wood density (WD). We examined the relationships between leaf reflectance and traits using partial least square regression (PLSR).</div><div>Our findings indicate that leaf reflectance accurately predicts variation in LMA (R<sup>2</sup> = 0.8), and reasonably estimates xylem water potential (R<sup>2</sup> = 0.51) and WD (R<sup>2</sup> = 0.52). However, P50 predictions were much less reliable (R<sup>2</sup> = 0.27), with water absorption bands greatly influencing the PLSR model. Leaf phenological strategy had little impact on the results.</div><div>These findings suggest that reflectance-based remote sensing could monitor water status and forest carbon dynamics through water potential and wood density, respectively. However, our case study applying the PLSR approach to hyperspectral canopy spectra to predict wood density revealed challenges to upscaling. Despite these limitations, remote sensing of forest hydraulic traits at scale could enhance our understanding of drought vulnerability and carbon dynamics in Amazonian forests, with significant implications for conservation.</div></div>","PeriodicalId":417,"journal":{"name":"Remote Sensing of Environment","volume":"318 ","pages":"Article 114562"},"PeriodicalIF":11.1,"publicationDate":"2024-12-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142820924","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0