Science of Remote Sensing最新文献_第6页

The methodology for upscaling surface albedo with consideration of surface heterogeneity 考虑表面非均质性的地表反照率升标方法

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-12-01 DOI: 10.1016/j.srs.2025.100340

Ziyu Wang, Hongmin Zhou, Jia Tang, Chen Li, Jinlin Qi, Bihong Yang, Ruojing Fang

Pixel scale land surface truth value is essential for satellite product validation. Due to the heavy field observation burden, it is difficult to obtain large-scale field data. In recent years, machine learning has been widely used in upscaling processes due to its significant advantages in dealing with complex and nonlinear problems. However, surface heterogeneity, which may lead to significant errors in the prediction results, was seldom taken into consideration in the current machine learning upscaling methods. To address this issue, this study proposed a land surface upscaling model which incorporates land surface heterogeneity. A hybrid coefficient of variation (CV) index was adopted to depict the surface heterogeneity, and the random forest (RF) model was applied to upscale single-site in-situ albedo to the coarse pixel scale. The upscaled results were evaluated with pixel-scale albedo reference data. A high accuracy with R² of 0.93 and RMSE of 0.026 was obtained. To better understand the performance of the proposed model, 20km × 20km homogeneous and heterogeneous regions were also selected. Results show that models that take surface heterogeneity into account can improve the accuracy of predictions. In regions with low heterogeneity, whether or not surface heterogeneity was taken into account had less impact on the prediction results. However, in regions with high heterogeneity, the model considering surface heterogeneity better captured the spatial and temporal variations in albedo. This study proposed an effective albedo upscaling model with consideration of land surface heterogeneity, which can also be applied to the scale transformation of other land surface parameters.

像元尺度的地表真值是卫星产品验证的关键。由于野外观测负担较大，难以获得大规模的野外数据。近年来，机器学习因其在处理复杂和非线性问题方面的显著优势而被广泛应用于升级过程中。然而，在目前的机器学习升级方法中，很少考虑到表面非均质性会导致预测结果出现较大误差。为了解决这一问题，本研究提出了一个考虑地表异质性的地表尺度升级模型。采用混合变异系数（CV）指数描述地表异质性，采用随机森林（RF）模型将单点原位反照率提升至粗像元尺度。升级后的结果用像素级反照率参考数据进行评价。准确度较高，R2为0.93，RMSE为0.026。为了更好地理解模型的性能，还选择了20km × 20km的均匀和非均匀区域。结果表明，考虑地表非均质性的模型可以提高预测的准确性。在非均质性较低的区域，是否考虑地表非均质性对预测结果的影响较小。而在非均质性较高的区域，考虑地表非均质性的模式能更好地捕捉反照率的时空变化。本文提出了一种考虑地表非均质性的有效反照率升尺度模型，该模型也可应用于其他地表参数的尺度转换。

{"title":"The methodology for upscaling surface albedo with consideration of surface heterogeneity","authors":"Ziyu Wang, Hongmin Zhou, Jia Tang, Chen Li, Jinlin Qi, Bihong Yang, Ruojing Fang","doi":"10.1016/j.srs.2025.100340","DOIUrl":"10.1016/j.srs.2025.100340","url":null,"abstract":"<div><div>Pixel scale land surface truth value is essential for satellite product validation. Due to the heavy field observation burden, it is difficult to obtain large-scale field data. In recent years, machine learning has been widely used in upscaling processes due to its significant advantages in dealing with complex and nonlinear problems. However, surface heterogeneity, which may lead to significant errors in the prediction results, was seldom taken into consideration in the current machine learning upscaling methods. To address this issue, this study proposed a land surface upscaling model which incorporates land surface heterogeneity. A hybrid coefficient of variation (CV) index was adopted to depict the surface heterogeneity, and the random forest (RF) model was applied to upscale single-site <em>in-situ</em> albedo to the coarse pixel scale. The upscaled results were evaluated with pixel-scale albedo reference data. A high accuracy with R<sup>2</sup> of 0.93 and RMSE of 0.026 was obtained. To better understand the performance of the proposed model, 20km × 20km homogeneous and heterogeneous regions were also selected. Results show that models that take surface heterogeneity into account can improve the accuracy of predictions. In regions with low heterogeneity, whether or not surface heterogeneity was taken into account had less impact on the prediction results. However, in regions with high heterogeneity, the model considering surface heterogeneity better captured the spatial and temporal variations in albedo. This study proposed an effective albedo upscaling model with consideration of land surface heterogeneity, which can also be applied to the scale transformation of other land surface parameters.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100340"},"PeriodicalIF":5.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145624120","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 review of crop yield estimation on pixel and field scales from remotely sensed data 基于遥感数据的像元和田的作物产量估算综述

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-27 DOI: 10.1016/j.srs.2025.100342

Fengjiao Zhang , Shunlin Liang , Han Ma , Wenyuan Li , Yongzhe Chen , Tao He , Feng Tian , Jianglei Xu , Husheng Fang , Hui Liang , Yichuan Ma , Aolin Jia , Yuxiang Zhang

Crop yield estimation over large regions can provide critical data support for regional agricultural management and global food security assessments. The previous reviews mainly focused on the technological advancements of methods in specific areas such as crop growth, data assimilation, and machine learning. No reviews have summarized the progress in all these areas, particularly at the pixel and field scales. This review comprehensively evaluates various methods for estimating global and regional crop yield from different remotely sensed data, particularly on the pixel and field scales, in the past two decades. All estimation methods are grouped into four categories: empirical statistical, light use efficiency (LUE), data assimilation, and machine learning. We also identify remaining challenges in data consistency, update frequency, and crop type coverage, particularly in data-scarce developing regions. This review provides valuable insights for researchers in the field of remotely sensed data-based crop yield estimation, enabling a deeper understanding of the current status of global and regional datasets, the characteristics and challenges of existing estimation methods, and future research directions.

大区域作物产量估算可为区域农业管理和全球粮食安全评估提供关键数据支持。之前的综述主要集中在作物生长、数据同化和机器学习等特定领域的技术进展。没有一篇综述总结了所有这些领域的进展，特别是在像素和场尺度上的进展。本文综合评价了过去二十年来利用不同遥感数据估算全球和区域作物产量的各种方法，特别是在像元和田的尺度上。所有的估计方法分为四类：经验统计、光利用效率（LUE）、数据同化和机器学习。我们还确定了在数据一致性、更新频率和作物类型覆盖方面仍然存在的挑战，特别是在数据匮乏的发展中地区。本文综述为基于遥感数据的作物产量估算提供了有价值的见解，使研究人员能够更深入地了解全球和区域数据集的现状，现有估算方法的特点和挑战，以及未来的研究方向。

{"title":"A review of crop yield estimation on pixel and field scales from remotely sensed data","authors":"Fengjiao Zhang , Shunlin Liang , Han Ma , Wenyuan Li , Yongzhe Chen , Tao He , Feng Tian , Jianglei Xu , Husheng Fang , Hui Liang , Yichuan Ma , Aolin Jia , Yuxiang Zhang","doi":"10.1016/j.srs.2025.100342","DOIUrl":"10.1016/j.srs.2025.100342","url":null,"abstract":"<div><div>Crop yield estimation over large regions can provide critical data support for regional agricultural management and global food security assessments. The previous reviews mainly focused on the technological advancements of methods in specific areas such as crop growth, data assimilation, and machine learning. No reviews have summarized the progress in all these areas, particularly at the pixel and field scales. This review comprehensively evaluates various methods for estimating global and regional crop yield from different remotely sensed data, particularly on the pixel and field scales, in the past two decades. All estimation methods are grouped into four categories: empirical statistical, light use efficiency (LUE), data assimilation, and machine learning. We also identify remaining challenges in data consistency, update frequency, and crop type coverage, particularly in data-scarce developing regions. This review provides valuable insights for researchers in the field of remotely sensed data-based crop yield estimation, enabling a deeper understanding of the current status of global and regional datasets, the characteristics and challenges of existing estimation methods, and future research directions.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100342"},"PeriodicalIF":5.2,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145939364","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

Four decades of remote sensing for monitoring terrestrial ecosystems: a global review and future challenges 遥感监测陆地生态系统的四十年：全球回顾和未来挑战

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-27 DOI: 10.1016/j.srs.2025.100341

Jose Manuel Álvarez-Martínez , Tijana Nikolić Lugonja , Alicia Valdés , Jorge González Le Barbier , Marta Pérez Suárez , Gonzalo Hernández Romero , Mirjana Radulović , Maja Knežević , Sonja Tarčak , Branko Brkljač , Bojana Bokić , Boris Radak , Andrijana Andrić , Miljana Marković , Sanja Brdar , Predrag Lugonja , Isidora Simović , Lori Giagnacovo , Borja Jiménez-Alfaro

Remote sensing (RS) has evolved from occasional mapping to continuous, indicator-based monitoring of terrestrial ecosystems. This review synthesizes four decades of global progress in RS to characterize natural and semi-natural ecosystems, examining how study purposes, sensor types and analytical methods have diversified from 1985 to 2025. A systematic literature review of 6856 publications (1567 selected) documents the transition from expert-based visual interpretation using aerial photography and early Landsat missions, to harmonized, AI-driven workflows that enable scalable and replicable ecosystem assessments. Advances in cloud computing, data cubes and open-access archives now allow wall-to-wall time series of analyses across regions and biomes. Yet, important challenges persist, including the underrepresentation of biodiversity-rich areas, limited in-situ calibration data and uncertainties related to phenological variability, image correction or temporal mosaicking pipelines. Building on case studies from a global perspective, we outline design principles for policy-ready ecosystem indicators traceable to raw observations, comparable through time and space, and aligned with biodiversity policy frameworks. Integrating multi-sensor data (optical, radar, LiDAR, thermal), standardized in-situ observations and artificial intelligence/machine learning algorithms, RS provides a robust pathway towards operational ecosystem accounting and large-scale functional mapping and monitoring, strengthening conservation planning and ecosystem management worldwide.

遥感（RS）已经从偶尔的制图发展到对陆地生态系统进行连续的、基于指标的监测。本文综合了40年来全球自然和半自然生态系统遥感研究的进展，考察了1985年至2025年研究目的、传感器类型和分析方法的变化。对6856份出版物（选定的1567份）的系统文献综述记录了从使用航空摄影和早期陆地卫星任务的基于专家的视觉解释到协调的、人工智能驱动的工作流程的转变，从而实现了可扩展和可复制的生态系统评估。云计算、数据立方体和开放存取档案的进步现在允许跨地区和生物群落进行墙到墙的时间序列分析。然而，重要的挑战仍然存在，包括生物多样性丰富地区的代表性不足，有限的原位校准数据以及与物候变异性、图像校正或时间镶嵌管道相关的不确定性。以全球视角的案例研究为基础，我们概述了可追溯原始观测、可在时间和空间上进行比较并与生物多样性政策框架保持一致的政策导向生态系统指标的设计原则。RS集成了多传感器数据（光学、雷达、激光雷达、热）、标准化的原位观测和人工智能/机器学习算法，为操作生态系统会计和大规模功能制图和监测提供了强大的途径，加强了全球范围内的保护规划和生态系统管理。

{"title":"Four decades of remote sensing for monitoring terrestrial ecosystems: a global review and future challenges","authors":"Jose Manuel Álvarez-Martínez , Tijana Nikolić Lugonja , Alicia Valdés , Jorge González Le Barbier , Marta Pérez Suárez , Gonzalo Hernández Romero , Mirjana Radulović , Maja Knežević , Sonja Tarčak , Branko Brkljač , Bojana Bokić , Boris Radak , Andrijana Andrić , Miljana Marković , Sanja Brdar , Predrag Lugonja , Isidora Simović , Lori Giagnacovo , Borja Jiménez-Alfaro","doi":"10.1016/j.srs.2025.100341","DOIUrl":"10.1016/j.srs.2025.100341","url":null,"abstract":"<div><div>Remote sensing (RS) has evolved from occasional mapping to continuous, indicator-based monitoring of terrestrial ecosystems. This review synthesizes four decades of global progress in RS to characterize natural and semi-natural ecosystems, examining how study purposes, sensor types and analytical methods have diversified from 1985 to 2025. A systematic literature review of 6856 publications (1567 selected) documents the transition from expert-based visual interpretation using aerial photography and early Landsat missions, to harmonized, AI-driven workflows that enable scalable and replicable ecosystem assessments. Advances in cloud computing, data cubes and open-access archives now allow wall-to-wall time series of analyses across regions and biomes. Yet, important challenges persist, including the underrepresentation of biodiversity-rich areas, limited in-situ calibration data and uncertainties related to phenological variability, image correction or temporal mosaicking pipelines. Building on case studies from a global perspective, we outline design principles for policy-ready ecosystem indicators traceable to raw observations, comparable through time and space, and aligned with biodiversity policy frameworks. Integrating multi-sensor data (optical, radar, LiDAR, thermal), standardized in-situ observations and artificial intelligence/machine learning algorithms, RS provides a robust pathway towards operational ecosystem accounting and large-scale functional mapping and monitoring, strengthening conservation planning and ecosystem management worldwide.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"13 ","pages":"Article 100341"},"PeriodicalIF":5.2,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145748183","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

Detecting utility-scale solar installations and associated land cover changes using spatiotemporal segmentation of Landsat imagery 利用陆地卫星图像的时空分割检测公用事业规模的太阳能装置和相关的土地覆盖变化

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-18 DOI: 10.1016/j.srs.2025.100337

Kangjoon Cho, Curtis E. Woodcock

As utility-scale solar development expands across temperate forest regions, concerns have emerged over its unintended environmental impacts, particularly land-use and land-cover change (LULCC) and associated impacts on carbon dynamics. This study develops and tests an object-based change detection framework that integrates Continuous Change Detection and Classification (CCDC) with Simple Non-Iterative Clustering (SNIC) to map utility-scale solar development and related forest disturbances in Massachusetts from 2005 to 2024. By combining temporal and spatial segmentation, this method maps the spatiotemporal footprint of utility-scale solar installations with associated deforestation and post-clearing dynamics. Our results show approximately 51 % of utility-scale solar installations occurred on previously forested land, with an additional 64 % of surrounding forest cleared relative to the directly deforested area. On average, each hectare of new solar development on forest land was associated with 1.66 ha of deforestation between 2005 and 2024. Temporal trends from CCDC models quantify post-clearing vegetation dynamics and microclimatic effects. Exploration of the microclimatic effects of solar installations using time series method indicates Land Surface Temperature (LST) increases exceeding +7.2 °C in solar developments with deforestation and Albedo reductions after solar panel installations, suggesting substantial shifts in site-level biophysical conditions. Normalized Difference Vegetation Index (NDVI) time series showed a positive slope following solar installation, indicating gradual vegetation regrowth. The results indicate that the newly developed CCDC-SNIC was successfully applied to Landsat time series for detecting utility-scale solar installations and their associated LULCC. This method can provide a foundation for integrating LULCC data with carbon modeling to quantify the net climate impact of utility-scale solar.

随着公用事业规模的太阳能开发在温带森林地区的扩展，人们开始关注其意想不到的环境影响，特别是土地利用和土地覆盖变化（LULCC）及其对碳动态的相关影响。本研究开发并测试了一个基于对象的变化检测框架，该框架将连续变化检测和分类（CCDC）与简单非迭代聚类（SNIC）相结合，用于绘制2005年至2024年马萨诸塞州公用事业规模的太阳能开发和相关森林干扰。通过结合时间和空间分割，该方法绘制了公用事业规模太阳能装置的时空足迹与相关的森林砍伐和清理后动态。我们的研究结果显示，大约51%的公用事业规模的太阳能装置发生在以前的森林土地上，相对于直接砍伐森林的地区，另外64%的周围森林被砍伐。2005年至2024年期间，平均每公顷林地上的新太阳能开发与1.66公顷的森林砍伐有关。CCDC模型的时间趋势量化了砍伐后植被动态和小气候效应。利用时间序列方法探索太阳能装置的小气候效应表明，在太阳能电池板安装后，随着森林砍伐和反照率的降低，太阳能发展的陆地表面温度（LST）增加超过+7.2°C，表明站点水平生物物理条件发生了实质性变化。归一化植被指数（NDVI）时间序列在太阳能安装后呈现正斜率，表明植被逐渐恢复。结果表明，新开发的CCDC-SNIC成功地应用于Landsat时间序列，用于检测公用事业规模的太阳能装置及其相关的LULCC。该方法可以为将LULCC数据与碳模型相结合以量化公用事业规模太阳能的净气候影响提供基础。

{"title":"Detecting utility-scale solar installations and associated land cover changes using spatiotemporal segmentation of Landsat imagery","authors":"Kangjoon Cho, Curtis E. Woodcock","doi":"10.1016/j.srs.2025.100337","DOIUrl":"10.1016/j.srs.2025.100337","url":null,"abstract":"<div><div>As utility-scale solar development expands across temperate forest regions, concerns have emerged over its unintended environmental impacts, particularly land-use and land-cover change (LULCC) and associated impacts on carbon dynamics. This study develops and tests an object-based change detection framework that integrates Continuous Change Detection and Classification (CCDC) with Simple Non-Iterative Clustering (SNIC) to map utility-scale solar development and related forest disturbances in Massachusetts from 2005 to 2024. By combining temporal and spatial segmentation, this method maps the spatiotemporal footprint of utility-scale solar installations with associated deforestation and post-clearing dynamics. Our results show approximately 51 % of utility-scale solar installations occurred on previously forested land, with an additional 64 % of surrounding forest cleared relative to the directly deforested area. On average, each hectare of new solar development on forest land was associated with 1.66 ha of deforestation between 2005 and 2024. Temporal trends from CCDC models quantify post-clearing vegetation dynamics and microclimatic effects. Exploration of the microclimatic effects of solar installations using time series method indicates Land Surface Temperature (LST) increases exceeding +7.2 °C in solar developments with deforestation and Albedo reductions after solar panel installations, suggesting substantial shifts in site-level biophysical conditions. Normalized Difference Vegetation Index (NDVI) time series showed a positive slope following solar installation, indicating gradual vegetation regrowth. The results indicate that the newly developed CCDC-SNIC was successfully applied to Landsat time series for detecting utility-scale solar installations and their associated LULCC. This method can provide a foundation for integrating LULCC data with carbon modeling to quantify the net climate impact of utility-scale solar.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100337"},"PeriodicalIF":5.2,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145579197","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

Hyperspectral indicators of vegetation vitality across scales: From trees to forests 跨尺度植被活力的高光谱指标：从树木到森林

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-17 DOI: 10.1016/j.srs.2025.100336

Dominic Fawcett , Arthur Gessler , Katrin Meusburger , Christian Ginzler , David N. Steger , Ansgar Kahmen , Petra D'Odorico

Remotely sensed vegetation indices (VIs) are used as indicators of vegetation vitality and warning signals that indicate stress responses at the leaf and canopy level. Novel hyperspectral sensors increase the number and sensitivity of VIs which can be acquired from space-borne observations with the potential to map the health of forests at large scales. Besides stress-induced impacts on foliage, these VIs are also sensitive to confounding structural effects, relating to canopy cover, crown architecture and shadow fraction variations. Here we use remote sensing data at three increasingly coarse spatial resolutions acquired by drone and aircraft from the peak of the growing season in 2023 and 2024 to assess how VIs resolve vitality-related differences between tree crowns of different species and functional types. We show that most VIs still clearly resolve variations between tree crowns when using high (1 m) resolution aircraft data compared to very high (0.1 m) resolution drone-based data, which better distinguishes sunlit leaves from shade and non-photosynthetic material (R² = 0.46–0.86). At landscape scale, we found that data resampled to moderate (30 m) spatial resolution to mimic satellite data appeared sensitive to spatial variations in available water capacity (AWC) for the photochemical reflectance indices (PRI), chlorophyll and water content VIs. These indices are also considerably influenced by vegetation structure and functional type variations, which must be taken into account for mixed forests when mapping vegetation health indicators and linking tree-level monitoring networks to remote sensing information.

利用遥感植被指数（VIs）作为植被活力指标和预警信号，反映叶片和冠层水平的胁迫响应。新型高光谱传感器增加了可从天基观测获得的可见光的数量和灵敏度，有可能在大尺度上绘制森林健康图。除了应力对叶片的影响外，这些VIs还对与冠层覆盖度、树冠结构和阴影分数变化有关的混合结构效应敏感。本文利用无人机和飞机在2023年和2024年生长季节高峰期获取的三种越来越粗糙的空间分辨率遥感数据，评估VIs如何解决不同物种和功能类型树冠之间的活力相关差异。我们发现，与非常高（0.1 m）分辨率的无人机数据相比，使用高（1 m）分辨率的飞机数据时，大多数VIs仍然清楚地分辨出树冠之间的差异，后者更好地区分了阳光照射下的叶子、遮荫和非光合物质（R2 = 0.46-0.86）。在景观尺度上，我们发现，在中等（30 m）空间分辨率下，模拟卫星数据的数据对有效水容量（AWC）、光化学反射指数（PRI）、叶绿素和含水量VIs的空间变化很敏感，这些指数也受到植被结构和功能类型变化的显著影响。在绘制植被健康指标和将树级监测网络与遥感信息联系起来时，混交林必须考虑到这一点。

{"title":"Hyperspectral indicators of vegetation vitality across scales: From trees to forests","authors":"Dominic Fawcett , Arthur Gessler , Katrin Meusburger , Christian Ginzler , David N. Steger , Ansgar Kahmen , Petra D'Odorico","doi":"10.1016/j.srs.2025.100336","DOIUrl":"10.1016/j.srs.2025.100336","url":null,"abstract":"<div><div>Remotely sensed vegetation indices (VIs) are used as indicators of vegetation vitality and warning signals that indicate stress responses at the leaf and canopy level. Novel hyperspectral sensors increase the number and sensitivity of VIs which can be acquired from space-borne observations with the potential to map the health of forests at large scales. Besides stress-induced impacts on foliage, these VIs are also sensitive to confounding structural effects, relating to canopy cover, crown architecture and shadow fraction variations. Here we use remote sensing data at three increasingly coarse spatial resolutions acquired by drone and aircraft from the peak of the growing season in 2023 and 2024 to assess how VIs resolve vitality-related differences between tree crowns of different species and functional types. We show that most VIs still clearly resolve variations between tree crowns when using high (1 m) resolution aircraft data compared to very high (0.1 m) resolution drone-based data, which better distinguishes sunlit leaves from shade and non-photosynthetic material (R<sup>2</sup> = 0.46–0.86). At landscape scale, we found that data resampled to moderate (30 m) spatial resolution to mimic satellite data appeared sensitive to spatial variations in available water capacity (AWC) for the photochemical reflectance indices (PRI), chlorophyll and water content VIs. These indices are also considerably influenced by vegetation structure and functional type variations, which must be taken into account for mixed forests when mapping vegetation health indicators and linking tree-level monitoring networks to remote sensing information.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100336"},"PeriodicalIF":5.2,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145579196","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

Temporally dense 100-m land surface temperature retrieval via attention-based super-resolution deep learning 基于注意力的超分辨率深度学习反演时间密集100米地表温度

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-14 DOI: 10.1016/j.srs.2025.100335

Taufiq Rashid, Di Tian

Land surface temperature (LST) is a key variable for governing surface water and energy exchanges and supports applications in hydrology, water management, climate monitoring, and agriculture. The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) offers LST observations at fine spatial resolution (70 m) but suffers from limited temporal coverage, cloud interference, and mission duration constraints. To overcome these limitations, we develop an Attention-based Super-Resolution deep Residual network (ASRRN) that generates 100-m LST estimates with dense temporal coverage using coarse-resolution MODIS LST and multiple auxiliary datasets. ASRRN integrates convolutional layers, residual learning, and attention mechanisms to enhance spatial feature extraction, while the model uncertainty is quantified using Monte Carlo dropout. We examine the contribution of auxiliary inputs, including Harmonized Landsat Sentinel (HLS) surface reflectance, Sentinel-1 SAR, and digital elevation model (DEM), and find that the configuration using MODIS LST with HLS (SWIR1, NDVI) and DEM yields performance comparable to ECOSTRESS. Cross-validation over 12 heterogeneous regions across the contiguous United States and two sites in Australia shows that ASRRN outperforms the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), achieving higher correlation (r = 0.984 vs. 0.898), lower RMSE (2.267 K vs. 5.764 K), and lower MAE (1.626 K vs. 4.193 K) against observations. It also outperforms the Super-Resolution Convolutional Neural Network (SRCNN) and the Enhanced Deep Super-Resolution network (EDSR). Overall, ASRRN enables generation of temporally dense, 100-m LST image time series without using fine-resolution thermal observations at estimation time, advancing high-resolution LST monitoring for various applications.

地表温度（LST）是控制地表水和能量交换的关键变量，支持在水文、水管理、气候监测和农业等领域的应用。生态系统星载空间站热辐射计实验（ECOSTRESS）提供精细空间分辨率（70米）的地表温度观测，但受到有限的时间覆盖、云干扰和任务持续时间的限制。为了克服这些限制，我们开发了一种基于注意力的超分辨率深度残差网络（ASRRN），该网络使用粗分辨率MODIS LST和多个辅助数据集生成具有密集时间覆盖的100米LST估计。ASRRN集成了卷积层、残差学习和注意机制来增强空间特征提取，同时使用蒙特卡罗dropout来量化模型的不确定性。我们研究了辅助输入的贡献，包括Harmonized Landsat Sentinel （HLS）表面反射率、Sentinel-1 SAR和数字高程模型（DEM），并发现使用MODIS LST与HLS （SWIR1、NDVI）和DEM的配置产生的性能与ECOSTRESS相当。在美国连续12个异质区域和澳大利亚两个地点的交叉验证表明，ASRRN优于Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM)，与观测值的相关性更高（r = 0.984比0.898），RMSE更低（2.267 K比5.764 K）， MAE更低（1.626 K比4.193 K）。它也优于超分辨率卷积神经网络（SRCNN）和增强型深度超分辨率网络（EDSR）。总的来说，ASRRN能够生成时间密度为100米的地表温度图像时间序列，而无需在估计时间使用精细分辨率的热观测，从而推进了各种应用的高分辨率地表温度监测。

{"title":"Temporally dense 100-m land surface temperature retrieval via attention-based super-resolution deep learning","authors":"Taufiq Rashid, Di Tian","doi":"10.1016/j.srs.2025.100335","DOIUrl":"10.1016/j.srs.2025.100335","url":null,"abstract":"<div><div>Land surface temperature (LST) is a key variable for governing surface water and energy exchanges and supports applications in hydrology, water management, climate monitoring, and agriculture. The ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS) offers LST observations at fine spatial resolution (70 m) but suffers from limited temporal coverage, cloud interference, and mission duration constraints. To overcome these limitations, we develop an Attention-based Super-Resolution deep Residual network (ASRRN) that generates 100-m LST estimates with dense temporal coverage using coarse-resolution MODIS LST and multiple auxiliary datasets. ASRRN integrates convolutional layers, residual learning, and attention mechanisms to enhance spatial feature extraction, while the model uncertainty is quantified using Monte Carlo dropout. We examine the contribution of auxiliary inputs, including Harmonized Landsat Sentinel (HLS) surface reflectance, Sentinel-1 SAR, and digital elevation model (DEM), and find that the configuration using MODIS LST with HLS (SWIR1, NDVI) and DEM yields performance comparable to ECOSTRESS. Cross-validation over 12 heterogeneous regions across the contiguous United States and two sites in Australia shows that ASRRN outperforms the Enhanced Spatial and Temporal Adaptive Reflectance Fusion Model (ESTARFM), achieving higher correlation (r = 0.984 vs. 0.898), lower RMSE (2.267 K vs. 5.764 K), and lower MAE (1.626 K vs. 4.193 K) against observations. It also outperforms the Super-Resolution Convolutional Neural Network (SRCNN) and the Enhanced Deep Super-Resolution network (EDSR). Overall, ASRRN enables generation of temporally dense, 100-m LST image time series without using fine-resolution thermal observations at estimation time, advancing high-resolution LST monitoring for various applications.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100335"},"PeriodicalIF":5.2,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145579221","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

An optimised land-use land-cover classification approach for general application in deserts and arid regions 一种适用于沙漠和干旱地区的最佳土地利用土地覆盖分类方法

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-12 DOI: 10.1016/j.srs.2025.100334

João Carlos Campos , André Vicente Liz , László Patkó , Ayman Abdulkarem , Lourens Van Essen , Magdy El-Bana , Ahmed Al-Ansari , Omar Al-Attas , José Carlos Brito

Land-use/land-cover (LULC) is a critical driver of ecosystem dynamics globally. Arid regions are particularly vulnerable to global change factors, and comprehensive LULC assessments are crucial for evaluating the environmental stability of these areas. Despite considerable developments of remote sensing (RS) data/products and classification algorithms, the available multi-scale maps fail to represent the complex heterogeneity of LULC in these areas. The major limitation resides not on the improvements in data resolutions or the complexity of algorithmic decisions, but rather on the lack of approaches prioritizing detailed categorization of LULC in arid regions. Therefore, we propose an integrative multi-classification approach of LULC based on RS techniques and in-situ data for general application in arid regions, using the north-western Saudi Arabia as pilot area. Using in-situ data (N = 7523) and a Landsat-8 time-series, we applied a supervised classification of non-dynamic classes representing regional geodiversity, combined with a clustering analysis of dynamic harmonic regression model coefficients to categorize ecologically dynamic classes. The map was obtained with high accuracy for non-dynamic classes (Kappa≈0.84; overall accuracy≈0.87; overall producer's accuracy≈0.86; overall user's accuracy≈0.89) and dynamic classes (combined overall accuracy≈0.76). The final map presents a total of 15 classes, considerably improving the available categorical resolution for the study area. The approach is transferable to other arid regions, having the potential to integrate other finer-scale RS data and classification algorithms. We urge for increased efforts in data collection and the implementation of approaches considering the prominent diversity of LULC in arid regions.

土地利用/土地覆盖（LULC）是全球生态系统动态的关键驱动因素。干旱区特别容易受到全球变化因素的影响，综合的LULC评估对于评价干旱区的环境稳定性至关重要。尽管遥感数据/产品和分类算法有了长足的发展，但现有的多比例尺地图无法反映这些地区土地利用变化的复杂异质性。主要的限制不在于数据分辨率的提高或算法决策的复杂性，而在于缺乏对干旱地区LULC进行优先详细分类的方法。基于此，本文以沙特阿拉伯西北部地区为试点，提出了一种基于遥感技术和原位数据的综合多分类方法，该方法在干旱区具有普遍应用价值。利用实测数据（N = 7523）和Landsat-8时间序列，对代表区域地质多样性的非动态类别进行了监督分类，并结合动态调和回归模型系数的聚类分析对生态动态类别进行了分类。对于非动态类（Kappa≈0.84；总体精度≈0.87；总体生产者精度≈0.86；总体用户精度≈0.89）和动态类（综合总体精度≈0.76），获得了较高的精度。最终的地图共有15个类别，大大提高了研究区域的分类分辨率。该方法可推广到其他干旱地区，具有整合其他更精细尺度遥感数据和分类算法的潜力。我们敦促在数据收集方面加大努力，并采取考虑到干旱地区土地利用变化显著多样性的方法。

{"title":"An optimised land-use land-cover classification approach for general application in deserts and arid regions","authors":"João Carlos Campos , André Vicente Liz , László Patkó , Ayman Abdulkarem , Lourens Van Essen , Magdy El-Bana , Ahmed Al-Ansari , Omar Al-Attas , José Carlos Brito","doi":"10.1016/j.srs.2025.100334","DOIUrl":"10.1016/j.srs.2025.100334","url":null,"abstract":"<div><div>Land-use/land-cover (LULC) is a critical driver of ecosystem dynamics globally. Arid regions are particularly vulnerable to global change factors, and comprehensive LULC assessments are crucial for evaluating the environmental stability of these areas. Despite considerable developments of remote sensing (RS) data/products and classification algorithms, the available multi-scale maps fail to represent the complex heterogeneity of LULC in these areas. The major limitation resides not on the improvements in data resolutions or the complexity of algorithmic decisions, but rather on the lack of approaches prioritizing detailed categorization of LULC in arid regions. Therefore, we propose an integrative multi-classification approach of LULC based on RS techniques and in-situ data for general application in arid regions, using the north-western Saudi Arabia as pilot area. Using in-situ data (N = 7523) and a Landsat-8 time-series, we applied a supervised classification of non-dynamic classes representing regional geodiversity, combined with a clustering analysis of dynamic harmonic regression model coefficients to categorize ecologically dynamic classes. The map was obtained with high accuracy for non-dynamic classes (Kappa≈0.84; overall accuracy≈0.87; overall producer's accuracy≈0.86; overall user's accuracy≈0.89) and dynamic classes (combined overall accuracy≈0.76). The final map presents a total of 15 classes, considerably improving the available categorical resolution for the study area. The approach is transferable to other arid regions, having the potential to integrate other finer-scale RS data and classification algorithms. We urge for increased efforts in data collection and the implementation of approaches considering the prominent diversity of LULC in arid regions.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100334"},"PeriodicalIF":5.2,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528079","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

Deep learning-based vegetation canopy height mapping with polarimetric SAR: Application of a Polarization Fusion U-Net in Gabon’s tropical forests 基于深度学习的极化SAR植被冠层高度制图：偏振融合U-Net在加蓬热带森林中的应用

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-12 DOI: 10.1016/j.srs.2025.100327

Rezaul Hasan Bhuiyan , Claudia Paris , Tiejun Wang , Mahdi Khodadadzadeh , Michael Schlund

Forests provide essential ecosystem services, including carbon storage and the conservation of biodiversity. This highlights the need for accurate and scalable methods to assess forest structure. Active remote sensing, particularly Synthetic Aperture Radar (SAR), has significant potential for estimating forest structure owing to its ability to penetrate vegetation layers and interact with different forest elements. In particular, L- and P-band SAR signals are able to penetrate canopy layers, making them suitable for retrieving forest structure information.

We present a novel approach utilizing full-polarimetric SAR backscatter data to estimate canopy height as relevant forest structure variable. To capture the non-linear complex relationship between the SAR data and the vegetation canopy height, we propose a Polarization Fusion U-Net (PF-Unet) designed to enhance canopy height estimation from SAR backscatter data by effectively utilizing multi-polarization channels (e.g., HH, HV, and VV). Specifically, the proposed model is tailored to the physical properties of the SAR backscatter, incorporating: (a) a polarization fusion layer, (b) attention gates applied to each layer in the decoder blocks, and (c) Exponential Linear Unit (ELU) activation and Huber loss functions. To assess the potential of the PF-Unet model, L- and P-band SAR data, collected over the tropical forests of Gabon, were separately used. The model was evaluated in a complex tropical forest environment and compared against traditional Machine Learning (ML) approaches (Random Forest (RF) and Light Gradient Boosting Machine (LGBM)) as well as the standard U-Net model. The PF-Unet consistently outperformed all the baselines for both SAR datasets. The PF-Unet model achieved an RMSE of 4.35 m (15.73%) for L-band and 4.43 m (15.95%) for P-band, which was an improvement over the U-Net model’s RMSE of 5.02 m (18.15%) and 4.59 m (16.53%), respectively. This suggests promising potential for enhanced canopy height estimation, which is particularly valuable for upcoming spaceborne missions like NISAR and BIOMASS.

森林提供基本的生态系统服务，包括碳储存和生物多样性保护。这突出表明需要准确和可扩展的方法来评估森林结构。主动遥感，特别是合成孔径雷达（SAR），由于能够穿透植被层并与不同的森林要素相互作用，在估计森林结构方面具有巨大的潜力。特别是L波段和p波段SAR信号能够穿透林冠层，适用于森林结构信息的检索。本文提出了一种利用全极化SAR后向散射数据估算林冠高度的新方法。为了捕捉SAR数据与植被冠层高度之间的非线性复杂关系，我们提出了一种偏振融合U-Net (PF-Unet)，旨在通过有效利用多极化通道（如HH、HV和VV）来增强SAR后向散射数据的冠层高度估计。具体来说，所提出的模型针对SAR后向散射的物理特性进行了定制，包括：(a)极化融合层，(b)应用于解码器块中每层的注意门，以及(c)指数线性单元（ELU）激活和Huber损失函数。为了评估PF-Unet模式的潜力，分别使用了加蓬热带森林上空收集的L波段和p波段SAR数据。该模型在复杂的热带森林环境中进行了评估，并与传统的机器学习（ML）方法（随机森林（RF）和光梯度增强机（LGBM））以及标准的U-Net模型进行了比较。对于两个SAR数据集，PF-Unet始终优于所有基线。PF-Unet模型在l波段和p波段的RMSE分别为4.35 m（15.73%）和4.43 m(15.95%)，高于U-Net模型的5.02 m（18.15%）和4.59 m（16.53%）。这表明增强冠层高度估计的潜力很大，这对即将到来的NISAR和生物质等星载任务特别有价值。

{"title":"Deep learning-based vegetation canopy height mapping with polarimetric SAR: Application of a Polarization Fusion U-Net in Gabon’s tropical forests","authors":"Rezaul Hasan Bhuiyan , Claudia Paris , Tiejun Wang , Mahdi Khodadadzadeh , Michael Schlund","doi":"10.1016/j.srs.2025.100327","DOIUrl":"10.1016/j.srs.2025.100327","url":null,"abstract":"<div><div>Forests provide essential ecosystem services, including carbon storage and the conservation of biodiversity. This highlights the need for accurate and scalable methods to assess forest structure. Active remote sensing, particularly Synthetic Aperture Radar (SAR), has significant potential for estimating forest structure owing to its ability to penetrate vegetation layers and interact with different forest elements. In particular, L- and P-band SAR signals are able to penetrate canopy layers, making them suitable for retrieving forest structure information.</div><div>We present a novel approach utilizing full-polarimetric SAR backscatter data to estimate canopy height as relevant forest structure variable. To capture the non-linear complex relationship between the SAR data and the vegetation canopy height, we propose a Polarization Fusion U-Net (PF-Unet) designed to enhance canopy height estimation from SAR backscatter data by effectively utilizing multi-polarization channels (e.g., HH, HV, and VV). Specifically, the proposed model is tailored to the physical properties of the SAR backscatter, incorporating: (a) a polarization fusion layer, (b) attention gates applied to each layer in the decoder blocks, and (c) Exponential Linear Unit (ELU) activation and Huber loss functions. To assess the potential of the PF-Unet model, L- and P-band SAR data, collected over the tropical forests of Gabon, were separately used. The model was evaluated in a complex tropical forest environment and compared against traditional Machine Learning (ML) approaches (Random Forest (RF) and Light Gradient Boosting Machine (LGBM)) as well as the standard U-Net model. The PF-Unet consistently outperformed all the baselines for both SAR datasets. The PF-Unet model achieved an RMSE of 4.35 m (15.73%) for L-band and 4.43 m (15.95%) for P-band, which was an improvement over the U-Net model’s RMSE of 5.02 m (18.15%) and 4.59 m (16.53%), respectively. This suggests promising potential for enhanced canopy height estimation, which is particularly valuable for upcoming spaceborne missions like NISAR and BIOMASS.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100327"},"PeriodicalIF":5.2,"publicationDate":"2025-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528075","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

Discrepancies in cropland mapping and their implications for biodiversity conservation in China 中国耕地填图差异及其对生物多样性保护的启示

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-11 DOI: 10.1016/j.srs.2025.100332

Cong Ou , Yaqun Liu

The increasing public availability of long-term remote sensing imagery, combined with advancements in algorithms and cloud computing, has enabled the development of large-scale cropland mapping datasets at national and global levels. These datasets are critical for assessing the ecological and environmental risks of agricultural production in the Anthropocene era. While previous studies have primarily focused on improving the accuracy of cropland mapping, the implications of dataset discrepancies for evaluating cropland-induced ecological impacts remain insufficiently explored. Here, we integrate seven publicly available cropland datasets covering China and systematically analyze their spatiotemporal discrepancies, along with associated uncertainties in assessing habitat and biodiversity impacts. Our analysis incorporates key conservation areas, protected zones, species richness maps, and Dynamic Habitat Indices time series data. The results indicate that although all datasets indicate a general stabilization of cropland area in China over recent decades, significant discrepancies persist—particularly at finer spatial scales and in ecologically sensitive regions. These discrepancies not only influence the diagnosis of spatial conflicts between cropland and habitat areas but may also result in an underestimation of the threats posed by cropland expansion to habitats of threatened species. Furthermore, regions with greater cropland mapping discrepancies tend to exhibit lower environmental stability. Our findings highlight the need for selecting regionally appropriate cropland datasets when assessing historical and current cropland distributions and their ecological consequences. This study also offers insights into how remote sensing technologies can be better leveraged for ecological and agricultural sustainability research.

长期遥感图像的日益公开，再加上算法和云计算的进步，使得能够在国家和全球两级开发大规模农田制图数据集。这些数据集对于评估人类世时期农业生产的生态和环境风险至关重要。虽然以前的研究主要集中在提高农田制图的准确性，但数据集差异对评估农田生态影响的影响仍然没有得到充分的探讨。在此，我们整合了覆盖中国的7个公开耕地数据集，并系统分析了它们的时空差异，以及评估栖息地和生物多样性影响的相关不确定性。我们的分析结合了重点保护区、保护区、物种丰富度图和动态栖息地指数时间序列数据。结果表明，尽管所有数据集都表明近几十年来中国耕地面积总体稳定，但显著差异仍然存在，特别是在更精细的空间尺度和生态敏感区域。这些差异不仅影响耕地和栖息地之间空间冲突的诊断，而且可能导致低估耕地扩张对受威胁物种栖息地构成的威胁。此外，耕地制图差异较大的地区往往表现出较低的环境稳定性。我们的发现强调了在评估历史和当前农田分布及其生态后果时选择适合区域的农田数据集的必要性。这项研究还为如何更好地利用遥感技术进行生态和农业可持续性研究提供了见解。

{"title":"Discrepancies in cropland mapping and their implications for biodiversity conservation in China","authors":"Cong Ou , Yaqun Liu","doi":"10.1016/j.srs.2025.100332","DOIUrl":"10.1016/j.srs.2025.100332","url":null,"abstract":"<div><div>The increasing public availability of long-term remote sensing imagery, combined with advancements in algorithms and cloud computing, has enabled the development of large-scale cropland mapping datasets at national and global levels. These datasets are critical for assessing the ecological and environmental risks of agricultural production in the Anthropocene era. While previous studies have primarily focused on improving the accuracy of cropland mapping, the implications of dataset discrepancies for evaluating cropland-induced ecological impacts remain insufficiently explored. Here, we integrate seven publicly available cropland datasets covering China and systematically analyze their spatiotemporal discrepancies, along with associated uncertainties in assessing habitat and biodiversity impacts. Our analysis incorporates key conservation areas, protected zones, species richness maps, and Dynamic Habitat Indices time series data. The results indicate that although all datasets indicate a general stabilization of cropland area in China over recent decades, significant discrepancies persist—particularly at finer spatial scales and in ecologically sensitive regions. These discrepancies not only influence the diagnosis of spatial conflicts between cropland and habitat areas but may also result in an underestimation of the threats posed by cropland expansion to habitats of threatened species. Furthermore, regions with greater cropland mapping discrepancies tend to exhibit lower environmental stability. Our findings highlight the need for selecting regionally appropriate cropland datasets when assessing historical and current cropland distributions and their ecological consequences. This study also offers insights into how remote sensing technologies can be better leveraged for ecological and agricultural sustainability research.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100332"},"PeriodicalIF":5.2,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528078","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

High-resolution remote sensing-driven water management in semi-arid basins: A CNN-Attention-SWAT fusion framework for the Fen River 半干旱流域高分辨率遥感驱动的水资源管理：汾河CNN-Attention-SWAT融合框架

IF 5.2 Q1 ENVIRONMENTAL SCIENCES

Science of Remote Sensing

Pub Date : 2025-11-11 DOI: 10.1016/j.srs.2025.100333

Jiawen Liu , Xianqi Zhang , Yang Yang , Kemei Fu , Kaimin Wang

The Fen River Basin (FRB), a critical ecological corridor in China's Yellow River Basin, faces escalating water-security challenges under climate change and intensive human activities. Pressures include concentrated precipitation patterns, severe agricultural non-point source pollution (contributing >60 % nitrogen loads), groundwater overdraft, and increasing river flow interruptions (67 days/year in 2020), demanding integrated solutions aligned with Sustainable Development Goal 6 (SDG6). We propose a physics-embedded deep learning (PIDL) paradigm with bidirectional coupling between mechanistic and data-driven engines: 1) SWAT-modeled soil water stress index (SWSI) and groundwater depth are embedded into CNN-Attention-BiLSTM inputs to enforce physical plausibility; 2) Deep learning prediction errors dynamically update SWAT parameters (e.g., SOL_K, CH_N2) via Bayesian inversion. NSGA-II multi-objective optimization generates management strategies, validated through Monte Carlo simulations and ecological feasibility checks. The coupled framework outperformed standalone models in spatio-temporal accuracy: Runoff prediction: R² = 0.94, RMSE = 0.12 mm/d (37 % improvement vs. unidirectional coupling); Pollution load error reduced by 14.3 % (hotspot identification accuracy: ±1.5 km); Ecological flow compliance reached 92 % (vs. 69 % baseline). NSGA-II-optimized strategies achieved synergistic benefits: drip irrigation (65 % coverage, 12 % groundwater reduction), vegetative buffers (50m width, 31 % nitrogen load reduction), and dynamic ecological flows (dry season: 15 m³/s; wet season: 25 m³/s). Monte Carlo confirmed robustness (±11 % fluctuation). As implemented by Shanxi Water Resources Department, the PIDL framework enables cross-scale water governance (18–25 % systemic efficiency gain), balancing allocation, pollution control, and ecological restoration. Its "monitor-simulate-optimize-validate" architecture provides a replicable pathway for SDG-oriented management in semi-arid basins.

作为黄河流域重要生态廊道的汾河流域，在气候变化和人类活动加剧的背景下，面临着日益严峻的水安全挑战。压力包括集中降水模式、严重的农业非点源污染（造成60%的氮负荷）、地下水超量以及河流中断次数增加（2020年为67天/年），需要符合可持续发展目标6 （SDG6）的综合解决方案。我们提出了一种物理嵌入深度学习（PIDL）范式，该范式具有机制和数据驱动引擎之间的双向耦合：1)swat模型的土壤水分胁迫指数（SWSI）和地下水深度嵌入到CNN-Attention-BiLSTM输入中，以增强物理合理性；2)深度学习预测误差通过贝叶斯反演动态更新SWAT参数（如SOL_K， CH_N2）。NSGA-II多目标优化生成管理策略，通过蒙特卡罗模拟和生态可行性验证。耦合框架在时空精度上优于独立模型：径流预测：R2 = 0.94, RMSE = 0.12 mm/d（比单向耦合提高37%）；污染负荷误差降低14.3%（热点识别精度：±1.5 km）；生态流量顺应性达到92%（基线为69%）。nsga - ii优化策略实现了协同效益：滴灌（覆盖率65%，地下水减少12%）、植被缓冲（50米宽，氮负荷减少31%）和动态生态流量（旱季：15 m3/s，雨季：25 m3/s）。蒙特卡罗验证了稳健性（±11%波动）。在山西省水利厅实施的PIDL框架中，实现了跨尺度水治理（18 - 25%的系统效率增益）、平衡配置、污染控制和生态修复。其“监测-模拟-优化-验证”架构为半干旱盆地的可持续发展目标管理提供了可复制的途径。

{"title":"High-resolution remote sensing-driven water management in semi-arid basins: A CNN-Attention-SWAT fusion framework for the Fen River","authors":"Jiawen Liu , Xianqi Zhang , Yang Yang , Kemei Fu , Kaimin Wang","doi":"10.1016/j.srs.2025.100333","DOIUrl":"10.1016/j.srs.2025.100333","url":null,"abstract":"<div><div>The Fen River Basin (FRB), a critical ecological corridor in China's Yellow River Basin, faces escalating water-security challenges under climate change and intensive human activities. Pressures include concentrated precipitation patterns, severe agricultural non-point source pollution (contributing >60 % nitrogen loads), groundwater overdraft, and increasing river flow interruptions (67 days/year in 2020), demanding integrated solutions aligned with Sustainable Development Goal 6 (SDG6). We propose a physics-embedded deep learning (PIDL) paradigm with bidirectional coupling between mechanistic and data-driven engines: 1) SWAT-modeled soil water stress index (SWSI) and groundwater depth are embedded into CNN-Attention-BiLSTM inputs to enforce physical plausibility; 2) Deep learning prediction errors dynamically update SWAT parameters (e.g., SOL_K, CH_N2) via Bayesian inversion. NSGA-II multi-objective optimization generates management strategies, validated through Monte Carlo simulations and ecological feasibility checks. The coupled framework outperformed standalone models in spatio-temporal accuracy: Runoff prediction: R<sup>2</sup> = 0.94, RMSE = 0.12 mm/d (37 % improvement vs. unidirectional coupling); Pollution load error reduced by 14.3 % (hotspot identification accuracy: ±1.5 km); Ecological flow compliance reached 92 % (vs. 69 % baseline). NSGA-II-optimized strategies achieved synergistic benefits: drip irrigation (65 % coverage, 12 % groundwater reduction), vegetative buffers (50m width, 31 % nitrogen load reduction), and dynamic ecological flows (dry season: 15 m<sup>3</sup>/s; wet season: 25 m<sup>3</sup>/s). Monte Carlo confirmed robustness (±11 % fluctuation). As implemented by Shanxi Water Resources Department, the PIDL framework enables cross-scale water governance (18–25 % systemic efficiency gain), balancing allocation, pollution control, and ecological restoration. Its \"monitor-simulate-optimize-validate\" architecture provides a replicable pathway for SDG-oriented management in semi-arid basins.</div></div>","PeriodicalId":101147,"journal":{"name":"Science of Remote Sensing","volume":"12 ","pages":"Article 100333"},"PeriodicalIF":5.2,"publicationDate":"2025-11-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145528077","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