International journal of applied earth observation and geoinformation : ITC journal最新文献_第7页

A rangeland management-oriented approach to map dry savanna − Woodland mosaics 以牧场管理为导向的干旱稀树草原--林地镶嵌地图绘制方法

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-14 DOI: 10.1016/j.jag.2024.104193

Vera De Cauwer , Marie-Pascale Colace , John Mendelsohn , Telmo Antonio , Cornelis Van Der Waal

Tropical savannas have a patchy vegetation structure and heterogeneous composition that complicates their mapping and management. Land managers need detailed vegetation information, especially as tropical savannas often support extensive ranching systems or wildlife-based tourism and face specific challenges such as bush thickening, drought, bushfires and, in Africa, browsing by large game. Since existing methods to map savanna vegetation mosaics rarely provide the resolution or speed required, this study aimed to characterise savanna vegetation with sufficient detail for management purposes and sufficient generalisation for the assessment of processes at a landscape level, using an easy, quick, and cost-efficient system. The study area is a semi-arid savanna in a small game reserve south of Etosha National Park in Namibia. A rapid field assessment focused on the woody vegetation and used the Bitterlich method. Indicator species analysis and MRPP tests resulted in five mixed woody vegetation classes. Random Forest was used to model vegetation composition, structure and woody cover. The highest accuracy was obtained for vegetation composition (77 %) and the lowest for vegetation cover (71 %) with similar accuracies at a resolution of 10 m compared to 30 m. The most important predictors were a radar mosaic (ALOS PALSAR HV) and Sentinel-2 data representing days in wet and dry seasons, with MSAVI₂ a more suitable vegetation index than NDVI. Other predictors such as land surface temperature during winter nights, geology, and distance to water points contributed to the models. The final vegetation map contains 10 classes based on woody vegetation composition and structure. The most dominant classes were Colophospermum mopane – Terminalia prunioides woodland (33 %) and bushland (18 %) with grassland only covering 2.5 %. The method described here was driven by management requirements and can be used for bush control monitoring, quantifying the carbon pool and carrying capacity. It combines an old field survey method with free state-of-the-art datasets and algorithms. The focus on woody vegetation minimises the dependence on the intermittent presence of grasses and herbs in semi-arid savannas.

热带稀树草原的植被结构错落有致，植被组成也不尽相同，这使其绘图和管理变得更加复杂。土地管理者需要详细的植被信息，尤其是热带稀树草原通常支持着广泛的牧场系统或以野生动物为基础的旅游业，并面临着灌木丛增厚、干旱、丛林火灾以及非洲大型动物啃食等特殊挑战。由于绘制热带稀树草原植被镶嵌图的现有方法很少能提供所需的分辨率或速度，因此本研究旨在利用一种简便、快速和具有成本效益的系统，为管理目的提供足够详细的热带稀树草原植被特征，并为景观层面的过程评估提供足够的概括性。研究区域是纳米比亚埃托沙国家公园南部一个小型野生动物保护区内的半干旱稀树草原。快速实地评估的重点是木本植被，采用的是比特利希方法。通过指标物种分析和 MRPP 测试，得出了五个混合木本植被等级。随机森林用于模拟植被组成、结构和林木覆盖率。最重要的预测因子是雷达镶嵌（ALOS PALSAR HV）和代表干湿季节天数的哨兵-2 数据，其中 MSAVI2 是比 NDVI 更合适的植被指数。冬季夜间的地表温度、地质和与水源点的距离等其他预测因素也对模型有所贡献。最终的植被图包含 10 个基于木本植被组成和结构的等级。最主要的等级是 Colophospermum mopane - Terminalia prunioides 林地（33%）和灌木林（18%），草地仅占 2.5%。本文所述方法由管理要求驱动，可用于灌木控制监测、量化碳库和承载能力。它将古老的实地调查方法与免费的最新数据集和算法相结合。对木本植被的关注最大程度地减少了对半干旱稀树草原中间歇存在的草和草本植物的依赖。

{"title":"A rangeland management-oriented approach to map dry savanna − Woodland mosaics","authors":"Vera De Cauwer , Marie-Pascale Colace , John Mendelsohn , Telmo Antonio , Cornelis Van Der Waal","doi":"10.1016/j.jag.2024.104193","DOIUrl":"10.1016/j.jag.2024.104193","url":null,"abstract":"<div><div>Tropical savannas have a patchy vegetation structure and heterogeneous composition that complicates their mapping and management. Land managers need detailed vegetation information, especially as tropical savannas often support extensive ranching systems or wildlife-based tourism and face specific challenges such as bush thickening, drought, bushfires and, in Africa, browsing by large game. Since existing methods to map savanna vegetation mosaics rarely provide the resolution or speed required, this study aimed to characterise savanna vegetation with sufficient detail for management purposes and sufficient generalisation for the assessment of processes at a landscape level, using an easy, quick, and cost-efficient system. The study area is a semi-arid savanna in a small game reserve south of Etosha National Park in Namibia. A rapid field assessment focused on the woody vegetation and used the Bitterlich method. Indicator species analysis and MRPP tests resulted in five mixed woody vegetation classes. Random Forest was used to model vegetation composition, structure and woody cover. The highest accuracy was obtained for vegetation composition (77 %) and the lowest for vegetation cover (71 %) with similar accuracies at a resolution of 10 m compared to 30 m. The most important predictors were a radar mosaic (ALOS PALSAR HV) and Sentinel-2 data representing days in wet and dry seasons, with MSAVI<sub>2</sub> a more suitable vegetation index than NDVI. Other predictors such as land surface temperature during winter nights, geology, and distance to water points contributed to the models. The final vegetation map contains 10 classes based on woody vegetation composition and structure. The most dominant classes were <em>Colophospermum mopane – Terminalia prunioides</em> woodland (33 %) and bushland (18 %) with grassland only covering 2.5 %. The method described here was driven by management requirements and can be used for bush control monitoring, quantifying the carbon pool and carrying capacity. It combines an old field survey method with free state-of-the-art datasets and algorithms. The focus on woody vegetation minimises the dependence on the intermittent presence of grasses and herbs in semi-arid savannas.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104193"},"PeriodicalIF":7.6,"publicationDate":"2024-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142432770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

How can geostatistics help us understand deep learning? An exploratory study in SAR-based aircraft detection 地质统计学如何帮助我们理解深度学习？基于合成孔径雷达的飞机探测探索研究

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-14 DOI: 10.1016/j.jag.2024.104185

Lifu Chen , Zhenhuan Fang , Jin Xing , Xingmin Cai

Deep Neural Networks (DNNs) have garnered significant attention across various research domains due to their impressive performance, particularly Convolutional Neural Networks (CNNs), known for their exceptional accuracy in image processing tasks. However, the opaque nature of DNNs has raised concerns about their trustworthiness, as users often cannot understand how the model arrives at its predictions or decisions. This lack of transparency is particularly problematic in critical fields such as healthcare, finance, and law, where the stakes are high. Consequently, there has been a surge in the development of explanation methods for DNNs. Typically, the effectiveness of these methods is assessed subjectively via human observation on the heatmaps or attribution maps generated by eXplanation AI (XAI) methods. In this paper, a novel GeoStatistics Explainable Artificial Intelligence (GSEAI) framework is proposed, which integrates spatial pattern analysis from Geostatistics with XAI algorithms to assess and compare XAI understandability. Global and local Moran’s I indices, commonly used to assess the spatial autocorrelation of geographic data, assist in comprehending the spatial distribution patterns of attribution maps produced by the XAI method, through measuring the levels of aggregation or dispersion. Interpreting and analyzing attribution maps by Moran’s I scattergram and LISA clustering maps provide an accurate global objective quantitative assessment of the spatial distribution of feature attribution and achieves a more understandable local interpretation. In this paper, we conduct experiments on aircraft detection in SAR images based on the widely used YOLOv5 network, and evaluate four mainstream XAI methods quantitatively and qualitatively. By using GSEAI to perform explanation analysis of the given DNN, we could gain more insights about the behavior of the network, to enhance the trustworthiness of DNN applications. To the best of our knowledge, this is the first time XAI has been integrated with geostatistical algorithms in SAR domain knowledge, which expands the analytical approaches of XAI and also promotes the development of XAI within SAR image analytics.

深度神经网络（DNN）因其令人印象深刻的性能而在各个研究领域备受关注，尤其是卷积神经网络（CNN），因其在图像处理任务中的卓越准确性而闻名。然而，卷积神经网络的不透明性引发了人们对其可信度的担忧，因为用户往往无法理解模型是如何得出预测或决策的。在医疗保健、金融和法律等利害关系重大的领域，这种缺乏透明度的问题尤为突出。因此，为 DNN 开发解释方法的热潮已经兴起。通常情况下，这些方法的有效性是通过人类对 eXplanation AI（XAI）方法生成的热图或归因图的观察进行主观评估的。本文提出了一个新颖的 GeoStatistics 可解释人工智能（GSEAI）框架，该框架将 Geostatistics 的空间模式分析与 XAI 算法相结合，以评估和比较 XAI 的可理解性。全局和局部莫兰 I 指数通常用于评估地理数据的空间自相关性，通过测量聚集或分散程度，有助于理解 XAI 方法生成的归因图的空间分布模式。通过 Moran's I 散点图和 LISA 聚类图来解释和分析归属图，可对特征归属的空间分布进行准确的全局客观定量评估，并实现更易于理解的局部解释。本文基于广泛使用的 YOLOv5 网络，对 SAR 图像中的飞机检测进行了实验，并对四种主流 XAI 方法进行了定量和定性评估。通过使用 GSEAI 对给定的 DNN 进行解释分析，我们可以获得更多关于网络行为的见解，从而提高 DNN 应用的可信度。据我们所知，这是 XAI 首次与合成孔径雷达领域知识中的地质统计算法相结合，拓展了 XAI 的分析方法，也促进了 XAI 在合成孔径雷达图像分析中的发展。

{"title":"How can geostatistics help us understand deep learning? An exploratory study in SAR-based aircraft detection","authors":"Lifu Chen , Zhenhuan Fang , Jin Xing , Xingmin Cai","doi":"10.1016/j.jag.2024.104185","DOIUrl":"10.1016/j.jag.2024.104185","url":null,"abstract":"<div><div>Deep Neural Networks (DNNs) have garnered significant attention across various research domains due to their impressive performance, particularly Convolutional Neural Networks (CNNs), known for their exceptional accuracy in image processing tasks. However, the opaque nature of DNNs has raised concerns about their trustworthiness, as users often cannot understand how the model arrives at its predictions or decisions. This lack of transparency is particularly problematic in critical fields such as healthcare, finance, and law, where the stakes are high. Consequently, there has been a surge in the development of explanation methods for DNNs. Typically, the effectiveness of these methods is assessed subjectively via human observation on the heatmaps or attribution maps generated by eXplanation AI (XAI) methods. In this paper, a novel GeoStatistics Explainable Artificial Intelligence (GSEAI) framework is proposed, which integrates spatial pattern analysis from Geostatistics with XAI algorithms to assess and compare XAI understandability. Global and local Moran’s I indices, commonly used to assess the spatial autocorrelation of geographic data, assist in comprehending the spatial distribution patterns of attribution maps produced by the XAI method, through measuring the levels of aggregation or dispersion. Interpreting and analyzing attribution maps by Moran’s I scattergram and LISA clustering maps provide an accurate global objective quantitative assessment of the spatial distribution of feature attribution and achieves a more understandable local interpretation. In this paper, we conduct experiments on aircraft detection in SAR images based on the widely used YOLOv5 network, and evaluate four mainstream XAI methods quantitatively and qualitatively. By using GSEAI to perform explanation analysis of the given DNN, we could gain more insights about the behavior of the network, to enhance the trustworthiness of DNN applications. To the best of our knowledge, this is the first time XAI has been integrated with geostatistical algorithms in SAR domain knowledge, which expands the analytical approaches of XAI and also promotes the development of XAI within SAR image analytics.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104185"},"PeriodicalIF":7.6,"publicationDate":"2024-10-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142433347","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Improving the observations of suspended sediment concentrations in rivers from Landsat to Sentinel-2 imagery 从大地遥感卫星到哨兵-2 图像改进河流悬浮泥沙浓度观测

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104209

Zhiqiang Qiu , Dong Liu , Nuoxiao Yan , Chen Yang , Panpan Chen , Chenxue Zhang , Hongtao Duan

Yellow River is famous for its exceptionally higher suspended sediment concentrations (SSC), displaying significant spatiotemporal heterogeneity across diverse sections. Although SSC monitoring of the Yellow River and some of its tributaries has been achieved using Landsat data, it remains unclear whether the inclusion of higher spatial resolution satellites can expand the spatiotemporal monitoring capabilities for the Yellow River and most of its tributaries. In this study, we employed Sentinel-2 imagery, offering superior spatiotemporal resolution, to develop a higher-accurate SSC model and quantitatively evaluated its potential to improve the spatiotemporal coverage of SSC monitoring compared to Landsat satellites. For the Yellow River in the Loess Plateau, the optimized Sentinel-2 model exhibited superior accuracy, achieving R² = 0.91, root mean square error of 728.76 mg/L, and unbiased percentage difference of 16.75%. Notably, distinct SSC distribution differences were observed across different rivers, indicating significant spatial heterogeneity (SSC: 0.58 – 3.01 × 10⁵ mg/L). Moreover, Sentinel-2 showed a significant increase in observation frequency and spatial coverage (204.08% and 107.15%, respectively) compared to Landsat. An additional 35.29% increase in observation frequency was achieved through the combined satellite observation method. Furthermore, based on river width statistics, we found that upgrading the spatial resolution from 10 m to 1 m enhanced the coverage of observable river segments in the Loess Plateau by approximately 47.96%, and by about 50.56% globally. This study established a crucial scientific foundation for integrating Sentinel-2 and Landsat, enabling finer-scale monitoring and management of river sediment.

黄河因其悬浮泥沙浓度（SSC）特别高而闻名，在不同河段显示出明显的时空异质性。虽然已经利用 Landsat 数据实现了对黄河及其部分支流的 SSC 监测，但目前仍不清楚纳入更高空间分辨率的卫星能否扩大对黄河及其大部分支流的时空监测能力。在本研究中，我们利用具有更高时空分辨率的哨兵-2 图像开发了更精确的 SSC 模型，并定量评估了其与 Landsat 卫星相比提高 SSC 监测时空覆盖率的潜力。对于黄土高原的黄河，优化后的 Sentinel-2 模型表现出更高的精度，R2 = 0.91，均方根误差为 728.76 mg/L，无偏百分比差为 16.75%。值得注意的是，不同河流的 SSC 分布存在明显差异，表明存在显著的空间异质性（SSC：0.58 - 3.01 × 105 mg/L）。此外，与 Landsat 相比，Sentinel-2 的观测频率和空间覆盖率都有显著提高（分别为 204.08% 和 107.15%）。通过联合卫星观测方法，观测频率又增加了 35.29%。此外，根据河宽统计，我们发现将空间分辨率从 10 米提高到 1 米，黄土高原可观测河段的覆盖率提高了约 47.96%，全球提高了约 50.56%。这项研究为整合哨兵-2 和大地遥感卫星奠定了重要的科学基础，从而能够对河流泥沙进行更精细的监测和管理。

{"title":"Improving the observations of suspended sediment concentrations in rivers from Landsat to Sentinel-2 imagery","authors":"Zhiqiang Qiu , Dong Liu , Nuoxiao Yan , Chen Yang , Panpan Chen , Chenxue Zhang , Hongtao Duan","doi":"10.1016/j.jag.2024.104209","DOIUrl":"10.1016/j.jag.2024.104209","url":null,"abstract":"<div><div>Yellow River is famous for its exceptionally higher suspended sediment concentrations (SSC), displaying significant spatiotemporal heterogeneity across diverse sections. Although SSC monitoring of the Yellow River and some of its tributaries has been achieved using Landsat data, it remains unclear whether the inclusion of higher spatial resolution satellites can expand the spatiotemporal monitoring capabilities for the Yellow River and most of its tributaries. In this study, we employed Sentinel-2 imagery, offering superior spatiotemporal resolution, to develop a higher-accurate SSC model and quantitatively evaluated its potential to improve the spatiotemporal coverage of SSC monitoring compared to Landsat satellites. For the Yellow River in the Loess Plateau, the optimized Sentinel-2 model exhibited superior accuracy, achieving <em>R<sup>2</sup></em> = 0.91, root mean square error of 728.76 mg/L, and unbiased percentage difference of 16.75%. Notably, distinct SSC distribution differences were observed across different rivers, indicating significant spatial heterogeneity (SSC: 0.58 – 3.01 × 10<sup>5</sup> mg/L). Moreover, Sentinel-2 showed a significant increase in observation frequency and spatial coverage (204.08% and 107.15%, respectively) compared to Landsat. An additional 35.29% increase in observation frequency was achieved through the combined satellite observation method. Furthermore, based on river width statistics, we found that upgrading the spatial resolution from 10 m to 1 m enhanced the coverage of observable river segments in the Loess Plateau by approximately 47.96%, and by about 50.56% globally. This study established a crucial scientific foundation for integrating Sentinel-2 and Landsat, enabling finer-scale monitoring and management of river sediment.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104209"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422887","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

UAV and field hyperspectral imaging for Sphagnum discrimination and vegetation modelling in Finnish aapa mires 无人飞行器和野外高光谱成像技术用于芬兰阿帕沼泽的泥炭藓鉴别和植被建模

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104201

Franziska Wolff , Sandra Lorenz , Pasi Korpelainen , Anette Eltner , Timo Kumpula

Detailed knowledge of vegetation patterns allows to evaluate mire ecosystems and their dynamics. The use of hyperspectral information has the benefits of exploring spectral characteristics of species and vegetation modelling. Our study employed multi-scale and multi-source hyperspectral imaging with a handheld camera in the field and an UAV (Unoccupied Aerial Vehicle) sensor covering the wavelengths of 400 – 1000 nm. Plot-level spectra acquired with a UAV and field spectra collected at 1 m height were combined to develop a spectral library for Sphagnum moss species. This library was then used to map dominant Sphagnum species in a Finnish Aapa mire complex using the Spectral Angle Mapper (SAM) classifier. Classification performance assessment was supported by calculating a water index from the UAV-information. Additionally, we examined the transferability of site-specific spectral libraries to an aapa mire with similar vegetation. The results showed little spectral variation in the plot spectrum between the sensors. A fusion of species- and plot-level libraries yielded the highest accuracy of 62 %. For both mires, there was a great variation among the class accuracies. Floating mosses had an accuracy of 86 %, followed by lawn-forming Sphagnum balticum with 77 %. For the test site, the latter species was mapped with an accuracy of 59 %. Red moss species achieved low accuracies of 45 % and 38 %, likely due to effects from sub-pixel and mixed-pixel effects of neighbouring graminoid species and the presence of litter. This might have also enhanced the contrast of adjacent pixels contributing to spectral alterations. Water table depth measurements and the water index revealed a hydrological preference for most species, with classification performance notably improving with higher water index values. We recommend collecting on-site hyperspectral information at varying hydrological circumstances to build a comprehensive spectral library for mire vegetation and modelling.

对植被模式的详细了解有助于评估沼泽生态系统及其动态。使用高光谱信息具有探索物种光谱特征和植被建模的好处。我们的研究采用了多尺度和多源高光谱成像技术，在野外使用手持相机，在无人机（UAV）上使用波长为 400-1000 纳米的传感器。利用无人飞行器采集的地块级光谱和在 1 米高处采集的野外光谱相结合，建立了一个斯帕格沼苔藓物种光谱库。然后，使用光谱角度绘图器（SAM）分类器，利用该库绘制芬兰阿帕沼泽群中的主要泥炭藓物种图。通过计算无人机信息中的水指数，对分类性能进行了评估。此外，我们还考察了特定地点光谱库在具有类似植被的阿帕沼泽中的可移植性。结果显示，传感器之间的地块光谱差异很小。物种库和地块库的融合产生了 62% 的最高准确率。在这两种沼泽中，不同类别的准确度差异很大。浮游苔藓的准确率为 86%，其次是草坪形成的 Sphagnum balticum，准确率为 77%。在测试地点，后一种苔藓的绘图准确率为 59%。红色苔藓物种的准确率较低，分别为 45% 和 38%，这可能是由于邻近禾本科物种的亚像素和混合像素效应以及垃圾的存在造成的。这也可能增强了相邻像素的对比度，导致光谱改变。地下水位深度测量和水指数显示了大多数物种的水文偏好，水指数值越高，分类效果越明显。我们建议在不同的水文条件下收集现场高光谱信息，为沼泽植被和建模建立一个全面的光谱库。

{"title":"UAV and field hyperspectral imaging for Sphagnum discrimination and vegetation modelling in Finnish aapa mires","authors":"Franziska Wolff , Sandra Lorenz , Pasi Korpelainen , Anette Eltner , Timo Kumpula","doi":"10.1016/j.jag.2024.104201","DOIUrl":"10.1016/j.jag.2024.104201","url":null,"abstract":"<div><div>Detailed knowledge of vegetation patterns allows to evaluate mire ecosystems and their dynamics. The use of hyperspectral information has the benefits of exploring spectral characteristics of species and vegetation modelling. Our study employed multi-scale and multi-source hyperspectral imaging with a handheld camera in the field and an UAV (Unoccupied Aerial Vehicle) sensor covering the wavelengths of 400 – 1000 nm. Plot-level spectra acquired with a UAV and field spectra collected at 1 m height were combined to develop a spectral library for <em>Sphagnum</em> moss species. This library was then used to map dominant <em>Sphagnum</em> species in a Finnish Aapa mire complex using the Spectral Angle Mapper (SAM) classifier. Classification performance assessment was supported by calculating a water index from the UAV-information. Additionally, we examined the transferability of site-specific spectral libraries to an aapa mire with similar vegetation. The results showed little spectral variation in the plot spectrum between the sensors. A fusion of species- and plot-level libraries yielded the highest accuracy of 62 %. For both mires, there was a great variation among the class accuracies. Floating mosses had an accuracy of 86 %, followed by lawn-forming <em>Sphagnum balticum</em> with 77 %. For the test site, the latter species was mapped with an accuracy of 59 %. Red moss species achieved low accuracies of 45 % and 38 %, likely due to effects from sub-pixel and mixed-pixel effects of neighbouring graminoid species and the presence of litter. This might have also enhanced the contrast of adjacent pixels contributing to spectral alterations. Water table depth measurements and the water index revealed a hydrological preference for most species, with classification performance notably improving with higher water index values. We recommend collecting on-site hyperspectral information at varying hydrological circumstances to build a comprehensive spectral library for mire vegetation and modelling.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104201"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422940","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Individual mapping of large polymorphic shrubs in high mountains using satellite images and deep learning 利用卫星图像和深度学习绘制高山大型多态灌木的个体分布图

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104191

Rohaifa Khaldi , Siham Tabik , Sergio Puertas-Ruiz , Julio Peñas de Giles , José Antonio Hódar Correa , Regino Zamora , Domingo Alcaraz Segura

Monitoring the distribution and size of long-living large shrubs, such as junipers, is crucial for assessing the long-term impacts of global change on high-mountain ecosystems. While deep learning models have shown remarkable success in object segmentation, adapting these models to detect shrub species with polymorphic nature remains challenging. In this research, we release a large dataset of individual shrub delineations on freely available satellite imagery and use an instance segmentation model to map all junipers over the treeline for an entire biosphere reserve (Sierra Nevada, Spain). To optimize performance, we introduced a novel dual data construction approach: using photo-interpreted (PI) data for model development and fieldwork (FW) data for validation. To account for the polymorphic nature of junipers during model evaluation, we developed a soft version of the Intersection over Union metric. Finally, we assessed the uncertainty of the resulting map in terms of canopy cover and density of shrubs per size class. Our model achieved an F1-score in shrub delineation of 87.87% on the PI data and 76.86% on the FW data. The R2 and RMSE of the observed versus predicted relationship were 0.63 and 6.67% for canopy cover, and 0.90 and 20.62 for shrub density. The greater density of larger shrubs in lower altitudes and smaller shrubs in higher altitudes observed in the model outputs was also present in the PI and FW data, suggesting an altitudinal uplift in the optimal performance of the species. This study demonstrates that deep learning applied on freely available high-resolution satellite imagery is useful to detect medium to large shrubs of high ecological value at the regional scale, which could be expanded to other high-mountains worldwide and to historical and fothcoming imagery.

监测杜松等长寿大型灌木的分布和大小对于评估全球变化对高山生态系统的长期影响至关重要。虽然深度学习模型在物体分割方面取得了显著的成功，但要使这些模型适用于检测具有多态性的灌木物种仍然具有挑战性。在这项研究中，我们在免费提供的卫星图像上发布了一个大型灌木个体划界数据集，并使用实例分割模型绘制了整个生物圈保护区（西班牙内华达山脉）树线上方的所有桧木。为了优化性能，我们引入了一种新颖的双重数据构建方法：使用照片解释（PI）数据进行模型开发，使用野外工作（FW）数据进行验证。为了在模型评估过程中考虑到桧木的多态性，我们开发了一个软版本的 "交集大于联合 "指标。最后，我们从树冠覆盖率和每个大小等级的灌木密度方面评估了所绘制地图的不确定性。我们的模型在灌木划分方面的 F1 分数在 PI 数据上达到了 87.87%，在 FW 数据上达到了 76.86%。树冠覆盖率的观测值与预测值关系的 R2 和 RMSE 分别为 0.63% 和 6.67%，灌木密度的观测值与预测值关系的 R2 和 RMSE 分别为 0.90% 和 20.62%。在模型输出中观察到的低海拔地区灌木密度较大，而高海拔地区灌木密度较小的现象也出现在 PI 和 FW 数据中，这表明物种的最佳表现在海拔上有所提升。这项研究表明，在免费提供的高分辨率卫星图像上应用深度学习，有助于在区域尺度上检测具有较高生态价值的大中型灌木。

{"title":"Individual mapping of large polymorphic shrubs in high mountains using satellite images and deep learning","authors":"Rohaifa Khaldi , Siham Tabik , Sergio Puertas-Ruiz , Julio Peñas de Giles , José Antonio Hódar Correa , Regino Zamora , Domingo Alcaraz Segura","doi":"10.1016/j.jag.2024.104191","DOIUrl":"10.1016/j.jag.2024.104191","url":null,"abstract":"<div><div>Monitoring the distribution and size of long-living large shrubs, such as junipers, is crucial for assessing the long-term impacts of global change on high-mountain ecosystems. While deep learning models have shown remarkable success in object segmentation, adapting these models to detect shrub species with polymorphic nature remains challenging. In this research, we release a large dataset of individual shrub delineations on freely available satellite imagery and use an instance segmentation model to map all junipers over the treeline for an entire biosphere reserve (Sierra Nevada, Spain). To optimize performance, we introduced a novel dual data construction approach: using photo-interpreted (PI) data for model development and fieldwork (FW) data for validation. To account for the polymorphic nature of junipers during model evaluation, we developed a soft version of the Intersection over Union metric. Finally, we assessed the uncertainty of the resulting map in terms of canopy cover and density of shrubs per size class. Our model achieved an F1-score in shrub delineation of 87.87% on the PI data and 76.86% on the FW data. The R2 and RMSE of the observed versus predicted relationship were 0.63 and 6.67% for canopy cover, and 0.90 and 20.62 for shrub density. The greater density of larger shrubs in lower altitudes and smaller shrubs in higher altitudes observed in the model outputs was also present in the PI and FW data, suggesting an altitudinal uplift in the optimal performance of the species. This study demonstrates that deep learning applied on freely available high-resolution satellite imagery is useful to detect medium to large shrubs of high ecological value at the regional scale, which could be expanded to other high-mountains worldwide and to historical and fothcoming imagery.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104191"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422941","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Solid-state LiDAR and IMU coupled urban road non-revisiting mapping 固态激光雷达和 IMU 耦合城市道路非重访测绘

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104207

Xiaolong Ma , Chun Liu , Akram Akbar , Yuanfan Qi , Xiaohang Shao , Yihong Qiao , Xuefei Shao

3D mapping provides highly accurate environmental data, which is essential for critical applications such as autonomous driving and urban emergency response. Light detection and ranging (LiDAR) sensors, particularly solid-state ones, play a pivotal role in spatial–temporal mapping by providing precise three-dimensional data of the environment, significantly enhancing remote sensing capabilities and adaptability to challenging environments compared to mechanical LiDAR systems. However, the limited field of view results in a sparse point cloud frame with few features, which poses challenges to feature matching, causes pose offset, and hinders spatial–temporal continuity, and further significant obstacle for existing vehicle-mounted mobile mapping methods. To address the above issues, we proposed a novel approach that integrating inertial measurement unit (IMU) with solid-state LiDAR. Specifically, it comprises two key modules: an initial localization mapping module, mitigating the limitations of solid-state LiDAR in positioning and mapping accuracy, and an attitude optimization mapping module utilizing real-time high-frequency IMU data to identify key frames for correcting initial attitudes and generating accurate 3D maps. The effectiveness of the method is validated through extensive experiments in complex community and high-speed urban road scenarios. Furthermore, our approach outperforms than the state-of-the-art techniques in test scenarios, achieving a significant 35% reduction in average absolute pose error and enhancing the robustness of vehicle-mounted mapping.

三维测绘可提供高精度的环境数据，对于自动驾驶和城市应急响应等关键应用至关重要。光探测与测距（LiDAR）传感器，尤其是固态传感器，通过提供精确的三维环境数据，在时空测绘中发挥着举足轻重的作用，与机械式 LiDAR 系统相比，大大提高了遥感能力和对挑战性环境的适应性。然而，有限的视场导致点云框架稀疏，特征较少，这给特征匹配带来了挑战，造成姿态偏移，阻碍了时空连续性，进一步成为现有车载移动测绘方法的重大障碍。为解决上述问题，我们提出了一种将惯性测量单元（IMU）与固态激光雷达相结合的新方法。具体来说，该方法包括两个关键模块：一个是初始定位测绘模块，用于缓解固态激光雷达在定位和测绘精度方面的局限性；另一个是姿态优化测绘模块，利用实时高频惯性测量单元数据识别关键帧，以纠正初始姿态并生成精确的三维地图。在复杂的社区和高速城市道路场景中进行的大量实验验证了该方法的有效性。此外，我们的方法在测试场景中的表现优于最先进的技术，将平均绝对姿态误差大幅降低了 35%，并增强了车载绘图的鲁棒性。

{"title":"Solid-state LiDAR and IMU coupled urban road non-revisiting mapping","authors":"Xiaolong Ma , Chun Liu , Akram Akbar , Yuanfan Qi , Xiaohang Shao , Yihong Qiao , Xuefei Shao","doi":"10.1016/j.jag.2024.104207","DOIUrl":"10.1016/j.jag.2024.104207","url":null,"abstract":"<div><div>3D mapping provides highly accurate environmental data, which is essential for critical applications such as autonomous driving and urban emergency response. Light detection and ranging (LiDAR) sensors, particularly solid-state ones, play a pivotal role in spatial–temporal mapping by providing precise three-dimensional data of the environment, significantly enhancing remote sensing capabilities and adaptability to challenging environments compared to mechanical LiDAR systems. However, the limited field of view results in a sparse point cloud frame with few features, which poses challenges to feature matching, causes pose offset, and hinders spatial–temporal continuity, and further significant obstacle for existing vehicle-mounted mobile mapping methods. To address the above issues, we proposed a novel approach that integrating inertial measurement unit (IMU) with solid-state LiDAR. Specifically, it comprises two key modules: an initial localization mapping module, mitigating the limitations of solid-state LiDAR in positioning and mapping accuracy, and an attitude optimization mapping module utilizing real-time high-frequency IMU data to identify key frames for correcting initial attitudes and generating accurate 3D maps. The effectiveness of the method is validated through extensive experiments in complex community and high-speed urban road scenarios. Furthermore, our approach outperforms than the state-of-the-art techniques in test scenarios, achieving a significant 35% reduction in average absolute pose error and enhancing the robustness of vehicle-mounted mapping.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104207"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A new attention-based deep metric model for crop type mapping in complex agricultural landscapes using multisource remote sensing data 利用多源遥感数据绘制复杂农业景观作物类型图的新型注意力深度度量模型

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104204

Yizhen Zheng , Wen Dong , ZhipingYang , Yihang Lu , Xin Zhang , Yanni Dong , Fengqing Sun

Accurate crop mapping is critical for agricultural decisions and food security. Despite the widespread use of machine learning and deep learning in remote sensing for crop classification, mapping crops in mountainous smallholder farming systems remains challenging. In particular, cloudy and rainy weather limits high-quality satellite imagery, potentially limiting the availability of reliable data for classification. Additionally, the substantial intraclass variability among multiple crops further impedes classification accuracy. In this context, this study sought to resolve these two issues by applying a hybrid approach that combines multisource remote sensing data and deep metric learning. For the first challenge, multisource remote sensing data, including Landsat-8, Sentinel-2 and Sentinel-1 data from the Google Earth Engine, were integrated to provide more comprehensive information on crop growth and differences. To address the second challenge, we proposed a 2D-CNN network enhanced by CBAM attention and an online hard example mining strategy. The network focuses on the channel-spatial information of crop samples and their surrounding pixels while promoting the convergence of similar crop samples within the latent feature space and enhancing the separation among different samples. This process is exemplified through a case study of crop mapping in Jiangjin District, Chongqing city, an area representing the typical mountain smallholder farming systems in Southwest China. Compared to six state-of-the-art methods, RF, SVM, XGBoost, ResNet18, and DMLOHM, our approach achieves the highest performance, with 93.99% overall accuracy, a kappa coefficient of 0.9253, and excellent F1 scores across numerous crop categories. The results of this study provide an effective solution for crop classification in complex mountainous regions and have promising potential for mapping under challenging natural conditions.

准确的作物测绘对于农业决策和粮食安全至关重要。尽管遥感技术中广泛使用机器学习和深度学习来进行作物分类，但绘制山区小农耕作系统中的作物图仍然具有挑战性。特别是，阴雨天气限制了高质量的卫星图像，可能会限制用于分类的可靠数据的可用性。此外，多种作物之间巨大的类内差异也进一步阻碍了分类的准确性。在这种情况下，本研究试图通过应用一种结合多源遥感数据和深度度量学习的混合方法来解决这两个问题。针对第一个挑战，我们整合了多源遥感数据，包括来自谷歌地球引擎的 Landsat-8、Sentinel-2 和 Sentinel-1 数据，以提供更全面的作物生长和差异信息。为应对第二个挑战，我们提出了一种由 CBAM 注意力和在线硬示例挖掘策略增强的 2D-CNN 网络。该网络关注作物样本及其周围像素的通道空间信息，同时促进潜在特征空间内相似作物样本的聚合，并增强不同样本之间的分离。重庆市江津区是中国西南地区典型的山区小农耕作制度的代表，该区的农作物制图案例研究就是这一过程的例证。与 RF、SVM、XGBoost、ResNet18 和 DMLOHM 等六种最先进的方法相比，我们的方法取得了最高的性能，总体准确率达 93.99%，卡帕系数为 0.9253，在众多作物类别中取得了优异的 F1 分数。本研究的结果为复杂山区的作物分类提供了一个有效的解决方案，并有望用于具有挑战性的自然条件下的绘图。

{"title":"A new attention-based deep metric model for crop type mapping in complex agricultural landscapes using multisource remote sensing data","authors":"Yizhen Zheng , Wen Dong , ZhipingYang , Yihang Lu , Xin Zhang , Yanni Dong , Fengqing Sun","doi":"10.1016/j.jag.2024.104204","DOIUrl":"10.1016/j.jag.2024.104204","url":null,"abstract":"<div><div>Accurate crop mapping is critical for agricultural decisions and food security. Despite the widespread use of machine learning and deep learning in remote sensing for crop classification, mapping crops in mountainous smallholder farming systems remains challenging. In particular, cloudy and rainy weather limits high-quality satellite imagery, potentially limiting the availability of reliable data for classification. Additionally, the substantial intraclass variability among multiple crops further impedes classification accuracy. In this context, this study sought to resolve these two issues by applying a hybrid approach that combines multisource remote sensing data and deep metric learning. For the first challenge, multisource remote sensing data, including Landsat-8, Sentinel-2 and Sentinel-1 data from the Google Earth Engine, were integrated to provide more comprehensive information on crop growth and differences. To address the second challenge, we proposed a 2D-CNN network enhanced by CBAM attention and an online hard example mining strategy. The network focuses on the channel-spatial information of crop samples and their surrounding pixels while promoting the convergence of similar crop samples within the latent feature space and enhancing the separation among different samples. This process is exemplified through a case study of crop mapping in Jiangjin District, Chongqing city, an area representing the typical mountain smallholder farming systems in Southwest China. Compared to six state-of-the-art methods, RF, SVM, XGBoost, ResNet18, and DMLOHM, our approach achieves the highest performance, with 93.99% overall accuracy, a kappa coefficient of 0.9253, and excellent F1 scores across numerous crop categories. The results of this study provide an effective solution for crop classification in complex mountainous regions and have promising potential for mapping under challenging natural conditions.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104204"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422885","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A TDFC-RNNs framework integrated temporal convolutional attention mechanism for InSAR surface deformation prediction: A case study in Beijing Plain 用于 InSAR 表面形变预测的 TDFC-RNNs 框架集成了时间卷积注意机制：北京平原案例研究

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104199

Sheng Yao , Changfeng Jing , Xu He , Yi He , Lifeng Zhang

The precise time series prediction method is the key technology for the monitoring and management of ground deformation. Current prediction methods mostly rely on independent sampling points for prediction, limiting the effective utilization of spatial features by the model, thereby affecting the overall spatial prediction accuracy, and it also restricts the prediction efficiency of the model. In response to the above-mentioned issues in previous research, this study proposes a Time Distributed Fully Connected (TDFC) Recurrent Neural Networks (RNNs) framework that integrates Temporal Convolutional Attention Mechanism (TCAM) for joint prediction of sampling points in time series Interferometric Synthetic Aperture Radar (InSAR) surface deformation data. Firstly, based on Sentinel-1A imagery over the Beijing Plain, the time series surface deformation data from May 2017 to April 2020 are obtained utilizing the Small Baseline Subset InSAR (SBAS-InSAR) technology. After data processing and production into a dataset, based on the TDFC-RNNs framework integrated with TCAM, five different RNN structures were used as prediction modules to construct time series prediction models for InSAR surface deformation. To investigate the effectiveness of the TCAM module on prediction performance, ablation experiments were conducted specifically targeting it. Furthermore, to explore the relative optimality choice of prediction modules under the current dataset and the compatibility of this framework with non-RNN structures, various other sequence models were selected as prediction modules. The predictive performance of the models constructed by this framework was compared in two aspects with benchmark methods, ablation models, and other exploratory models. This included evaluating the predictive results of the test set using various metrics and analyzing the trends in numerical characteristics of the predicted results for the next 60 time steps (720 days). The comprehensive comparison results indicate that the model constructed by this framework outperforms other methods or models in terms of overall performance across various evaluation metrics. At the same time, the future predicted results exhibit more reliable numerical characteristics, aligning well with the developmental trends of surface deformation. This suggests that the above-mentioned models demonstrate favorable predictive capabilities for time series InSAR surface deformation. Such results can be instrumental in intuitively assessing the overall situation of surface deformation in the study area, promptly identifying risks, and swiftly implementing measures to address potential hazards.

精确的时间序列预测方法是地面变形监测和管理的关键技术。目前的预测方法大多依靠独立采样点进行预测，限制了模型对空间特征的有效利用，从而影响了整体空间预测精度，也制约了模型的预测效率。针对前人研究中存在的上述问题，本研究提出了一种时间分布全连接（TDFC）循环神经网络（RNNs）框架，该框架集成了时序卷积注意机制（TCAM），用于时间序列干涉合成孔径雷达（InSAR）地表形变数据中采样点的联合预测。首先，基于北京平原上空的哨兵-1A 图像，利用小基线子集 InSAR（SBAS-InSAR）技术获得 2017 年 5 月至 2020 年 4 月的时间序列地表形变数据。数据处理并生成数据集后，基于与 TCAM 集成的 TDFC-RNNs 框架，使用五种不同的 RNN 结构作为预测模块，构建 InSAR 地面形变时间序列预测模型。为了研究 TCAM 模块对预测性能的影响，专门针对该模块进行了烧蚀实验。此外，为了探索在当前数据集下预测模块的相对最优选择以及该框架与非 RNN 结构的兼容性，还选择了其他各种序列模型作为预测模块。本框架构建的模型的预测性能与基准方法、消融模型和其他探索性模型进行了两方面的比较。这包括使用各种指标评估测试集的预测结果，以及分析接下来 60 个时间步长（720 天）的预测结果数值特征趋势。综合比较结果表明，该框架构建的模型在各种评价指标上的整体性能均优于其他方法或模型。同时，未来预测结果表现出更可靠的数值特征，与地表变形的发展趋势非常吻合。这表明上述模型对时间序列 InSAR 表面形变具有良好的预测能力。这些结果有助于直观地评估研究区域地表变形的总体情况，及时识别风险，并迅速采取措施应对潜在的危害。

{"title":"A TDFC-RNNs framework integrated temporal convolutional attention mechanism for InSAR surface deformation prediction: A case study in Beijing Plain","authors":"Sheng Yao , Changfeng Jing , Xu He , Yi He , Lifeng Zhang","doi":"10.1016/j.jag.2024.104199","DOIUrl":"10.1016/j.jag.2024.104199","url":null,"abstract":"<div><div>The precise time series prediction method is the key technology for the monitoring and management of ground deformation. Current prediction methods mostly rely on independent sampling points for prediction, limiting the effective utilization of spatial features by the model, thereby affecting the overall spatial prediction accuracy, and it also restricts the prediction efficiency of the model. In response to the above-mentioned issues in previous research, this study proposes a Time Distributed Fully Connected (TDFC) Recurrent Neural Networks (RNNs) framework that integrates Temporal Convolutional Attention Mechanism (TCAM) for joint prediction of sampling points in time series Interferometric Synthetic Aperture Radar (InSAR) surface deformation data. Firstly, based on Sentinel-1A imagery over the Beijing Plain, the time series surface deformation data from May 2017 to April 2020 are obtained utilizing the Small Baseline Subset InSAR (SBAS-InSAR) technology. After data processing and production into a dataset, based on the TDFC-RNNs framework integrated with TCAM, five different RNN structures were used as prediction modules to construct time series prediction models for InSAR surface deformation. To investigate the effectiveness of the TCAM module on prediction performance, ablation experiments were conducted specifically targeting it. Furthermore, to explore the relative optimality choice of prediction modules under the current dataset and the compatibility of this framework with non-RNN structures, various other sequence models were selected as prediction modules. The predictive performance of the models constructed by this framework was compared in two aspects with benchmark methods, ablation models, and other exploratory models. This included evaluating the predictive results of the test set using various metrics and analyzing the trends in numerical characteristics of the predicted results for the next 60 time steps (720 days). The comprehensive comparison results indicate that the model constructed by this framework outperforms other methods or models in terms of overall performance across various evaluation metrics. At the same time, the future predicted results exhibit more reliable numerical characteristics, aligning well with the developmental trends of surface deformation. This suggests that the above-mentioned models demonstrate favorable predictive capabilities for time series InSAR surface deformation. Such results can be instrumental in intuitively assessing the overall situation of surface deformation in the study area, promptly identifying risks, and swiftly implementing measures to address potential hazards.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104199"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Immediate assessment of forest fire using a novel vegetation index and machine learning based on multi-platform, high temporal resolution remote sensing images 利用基于多平台、高时间分辨率遥感图像的新型植被指数和机器学习对森林火灾进行即时评估

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104210

Hanqiu Xu , Jiahui Chen , Guojin He , Zhongli Lin , Yafen Bai , Mengjie Ren , Hao Zhang , Huimin Yin , Fenfen Liu

Forest fires pose a significant threat to ecosystems, biodiversity, and human settlements, necessitating accurate and timely detection of burned areas for post-fire management. This study focused on the immediate assessment of a recent major forest fire that occurred on March 15, 2024, in southwestern China. We comprehensively utilized high temporal resolution MODIS and Black Marble nighttime light images to monitor the fire’s development and introduced a novel method for detecting burned forest areas using a new Shadow-Enhanced Vegetation Index (SEVI) coupling with a machine learning technique. The SEVI effectively enhances the vegetation index (VI) values on shaded slopes and hence reduces the VI disparity between shaded and sunlit areas, which is critical for accurately extracting fire scars in such terrain. While SEVI primarily identifies burned forest areas, the Random Forest (RF) technique detects all burned areas, including both forested and non-forested regions. Consequently, the total burned area of the Yajiang forest fire was estimated at 23,588 ha, with the burned forest area covering 19,266 ha. The combination of SEVI and RF algorithms provided a comprehensive and efficient tool for identifying burned areas. Additionally, our study employed the Remote Sensing-based Ecological Index (RSEI) to assess the ecological impact of the fire on the region, uncovering an immediate 15 % decline in regional ecological conditions following the fire. The usage of RSEI has the potential to quantitatively understand ecological responses to the fire. The findings achieved in this study underscore the significance of precise fire-burned area extraction techniques for enhancing forest fire management and ecosystem recovery strategies, while also highlighting the broader ecological implications of such events.

森林火灾对生态系统、生物多样性和人类居住区构成重大威胁，因此需要准确及时地探测火灾区域，以便进行火后管理。本研究的重点是对 2024 年 3 月 15 日发生在中国西南部的一场重大森林火灾进行即时评估。我们综合利用了高时间分辨率的 MODIS 和黑云母夜间光照图像来监测火灾的发展，并引入了一种新的方法，即利用新的阴影增强植被指数（SEVI）与机器学习技术相结合来检测烧毁林区。SEVI 可有效增强阴影斜坡上的植被指数（VI）值，从而缩小阴影和阳光照射区域之间的植被指数差异，这对于在此类地形中准确提取火烧痕至关重要。SEVI 主要识别烧毁的森林区域，而随机森林（RF）技术则检测所有烧毁区域，包括森林和非森林区域。因此，雅江森林火灾的总烧毁面积估计为 23,588 公顷，其中烧毁森林面积为 19,266 公顷。SEVI 算法和射频算法的结合为识别烧毁区域提供了一个全面、高效的工具。此外，我们的研究还采用了基于遥感的生态指数（RSEI）来评估火灾对该地区的生态影响，发现火灾发生后，该地区的生态条件立即下降了 15%。使用 RSEI 有可能从数量上了解火灾对生态的影响。本研究的发现强调了精确的火灾燃烧区提取技术对加强森林火灾管理和生态系统恢复战略的重要意义，同时也突出了此类事件对生态的广泛影响。

{"title":"Immediate assessment of forest fire using a novel vegetation index and machine learning based on multi-platform, high temporal resolution remote sensing images","authors":"Hanqiu Xu , Jiahui Chen , Guojin He , Zhongli Lin , Yafen Bai , Mengjie Ren , Hao Zhang , Huimin Yin , Fenfen Liu","doi":"10.1016/j.jag.2024.104210","DOIUrl":"10.1016/j.jag.2024.104210","url":null,"abstract":"<div><div>Forest fires pose a significant threat to ecosystems, biodiversity, and human settlements, necessitating accurate and timely detection of burned areas for post-fire management. This study focused on the immediate assessment of a recent major forest fire that occurred on March 15, 2024, in southwestern China. We comprehensively utilized high temporal resolution MODIS and Black Marble nighttime light images to monitor the fire’s development and introduced a novel method for detecting burned forest areas using a new Shadow-Enhanced Vegetation Index (SEVI) coupling with a machine learning technique. The SEVI effectively enhances the vegetation index (VI) values on shaded slopes and hence reduces the VI disparity between shaded and sunlit areas, which is critical for accurately extracting fire scars in such terrain. While SEVI primarily identifies burned forest areas, the Random Forest (RF) technique detects all burned areas, including both forested and non-forested regions. Consequently, the total burned area of the Yajiang forest fire was estimated at 23,588 ha, with the burned forest area covering 19,266 ha. The combination of SEVI and RF algorithms provided a comprehensive and efficient tool for identifying burned areas. Additionally, our study employed the Remote Sensing-based Ecological Index (RSEI) to assess the ecological impact of the fire on the region, uncovering an immediate 15 % decline in regional ecological conditions following the fire. The usage of RSEI has the potential to quantitatively understand ecological responses to the fire. The findings achieved in this study underscore the significance of precise fire-burned area extraction techniques for enhancing forest fire management and ecosystem recovery strategies, while also highlighting the broader ecological implications of such events.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104210"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422891","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Quantification of the spatiotemporal dynamics of diurnal fog and low stratus occurrence in subtropical montane cloud forests using Himawari-8 imagery and topographic attributes 利用 Himawari-8 图像和地形属性量化亚热带山地云雾林中昼雾和低层气发生的时空动态变化

IF 7.6 Q1 REMOTE SENSING

International journal of applied earth observation and geoinformation : ITC journal

Pub Date : 2024-10-11 DOI: 10.1016/j.jag.2024.104212

Jie-Yun Chong , Min-Hui Lo , Cho-ying Huang

Montane cloud forests (MCFs) feature frequent, wind-driven cloud bands (fog and low stratus [FLS]), providing crucial moisture to the ecosystems. Elevated temperatures may displace FLS, impacting MCFs significantly. To evaluate the consequences, quantifying FLS occurrences is vital. In this study, we employed “RANdom forest GEneRator” (Ranger), an advanced machine learning algorithm, to detect diurnal (07:00–17:00) FLS (dFLS) occurrence from 2018 to 2021 in MCFs in northeast Taiwan using 31 variables, including the visible and infrared bands of the Advanced Himawari Imager onboard Himawari-8, pixel solar azimuth and zenith angles, band differences, the Normalized Difference Vegetation Index (NDVI) and topographic attributes. We applied simple (lumping all data) and three-mode (sunrise/sunset, cloudy and clear sky) models to predict dFLS occurrence. We randomly selected 80 % of the data for model development and the rest for validation by referring to four ground dFLS observation stations across an elevation range of 1151–1811 m a.s.l with 53,358 diurnal time-lapse photographs. We found that it was possible to detect dFLS occurrence in MCFs using both simple and three-mode models regardless of the weather conditions (F1 ≥ 0.864, accuracy ≥ 0.905 and the Matthews correlation coefficient ≥ 0.786); the performance of the simple model was slightly better. The NDVI was more important than other variables in both models. This study demonstrates that Ranger may be able to detect dFLS in MCFs solely using a comprehensive array of satellite features insensitive to varying atmospheric conditions and terrain effects, permitting systematic monitoring of dFLS over vast regions.

山地云雾林（MCFs）的特点是经常出现由风驱动的云带（雾和低层云 [FLS]），为生态系统提供重要的水分。气温升高可能会驱散雾和低层云，从而对山地云雾林造成严重影响。要评估其后果，量化雾和低层云的出现至关重要。在本研究中，我们采用先进的机器学习算法 "RANdom forest GEneRator"（Ranger），利用 31 个变量，包括 "向日葵-8 "号上的 "先进向日葵成像仪 "的可见光和红外波段、像素太阳方位角和天顶角、波段差异、归一化植被指数（NDVI）和地形属性，来检测 2018 年至 2021 年台湾东北部 MCF 的昼夜（07:00-17:00）FLS（dFLS）发生情况。我们采用简单模型（将所有数据合并）和三模式模型（日出/日落、阴天和晴天）来预测 dFLS 的发生。我们随机选取了 80% 的数据用于模型开发，其余数据用于验证，参考了海拔 1151-1811 米范围内的四个地面 dFLS 观测站，以及 53,358 张昼夜延时照片。我们发现，无论天气条件如何，使用简单模式和三模式模型都能检测到 MCF 中出现的 dFLS（F1 ≥ 0.864，准确度≥ 0.905，马修斯相关系数≥ 0.786）；简单模式的性能稍好。在两个模型中，NDVI 都比其他变量更重要。这项研究表明，护林员可以仅利用对不同大气条件和地形影响不敏感的综合卫星特征阵列来检测微卷叶螟，从而对广大地区的微卷叶螟进行系统监测。

{"title":"Quantification of the spatiotemporal dynamics of diurnal fog and low stratus occurrence in subtropical montane cloud forests using Himawari-8 imagery and topographic attributes","authors":"Jie-Yun Chong , Min-Hui Lo , Cho-ying Huang","doi":"10.1016/j.jag.2024.104212","DOIUrl":"10.1016/j.jag.2024.104212","url":null,"abstract":"<div><div>Montane cloud forests (MCFs) feature frequent, wind-driven cloud bands (fog and low stratus [FLS]), providing crucial moisture to the ecosystems. Elevated temperatures may displace FLS, impacting MCFs significantly. To evaluate the consequences, quantifying FLS occurrences is vital. In this study, we employed “RANdom forest GEneRator” (Ranger), an advanced machine learning algorithm, to detect diurnal (07:00–17:00) FLS (dFLS) occurrence from 2018 to 2021 in MCFs in northeast Taiwan using 31 variables, including the visible and infrared bands of the Advanced Himawari Imager onboard Himawari-8, pixel solar azimuth and zenith angles, band differences, the Normalized Difference Vegetation Index (NDVI) and topographic attributes. We applied simple (lumping all data) and three-mode (sunrise/sunset, cloudy and clear sky) models to predict dFLS occurrence. We randomly selected 80 % of the data for model development and the rest for validation by referring to four ground dFLS observation stations across an elevation range of 1151–1811 m a.s.l with 53,358 diurnal time-lapse photographs. We found that it was possible to detect dFLS occurrence in MCFs using both simple and three-mode models regardless of the weather conditions (F1 ≥ 0.864, accuracy ≥ 0.905 and the Matthews correlation coefficient ≥ 0.786); the performance of the simple model was slightly better. The NDVI was more important than other variables in both models. This study demonstrates that Ranger may be able to detect dFLS in MCFs solely using a comprehensive array of satellite features insensitive to varying atmospheric conditions and terrain effects, permitting systematic monitoring of dFLS over vast regions.</div></div>","PeriodicalId":73423,"journal":{"name":"International journal of applied earth observation and geoinformation : ITC journal","volume":"134 ","pages":"Article 104212"},"PeriodicalIF":7.6,"publicationDate":"2024-10-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"142422894","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0