Ecological Informatics最新文献_第6页

Assessing the impact of land use and land cover on predicting in-stream total phosphorus using GIS and machine learning models 利用GIS和机器学习模型评估土地利用和土地覆盖对预测河川总磷的影响

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-07 DOI: 10.1016/j.ecoinf.2026.103598

Hye Won Lee , Min Kim , Baehyun Min , Jung Hyun Choi

This study presents a spatially explicit framework for predicting in-stream total phosphorus (TP) concentrations by combining a convolutional neural network (CNN) with explainable AI techniques. A CNN–dense neural network (CNN–DNN) model was trained on five-year average in-stream TP data from 608 monitoring stations across the four major river basins of the Republic of Korea, using 22 detailed land use and land cover (LULC) classes, a digital elevation model (DEM), and slope. The model outperformed proportion-only regression, highlighting the added value of spatial LULC representation. Unlike previous TP prediction studies that relied primarily on proportional LULC metrics, our approach preserves the spatial arrangement of LULC patches through CNN-based feature extraction and integrates SHapley Additive exPlanations (SHAP) and Integrated Gradients (IG) to provide spatially explicit attribution of patch-level influences. This enables detection of near-stream clustering effects and landscape configurations that cannot be captured by proportion-only models. Model interpretation with SHAP revealed that residential zones generally reduced TP, while commercial and public utility areas increased it when clustered near streams. Paddy fields consistently exerted the strongest positive influence in agricultural watersheds, modulated by slope and hydrologic connectivity. Forests largely buffered TP, though mixed stands and adjacent industrial parcels occasionally increased nutrient loads. To enhance robustness, IG analysis with three baselines revealed consistent attribution patterns that reinforced SHAP findings. Our results reveal that spatial watershed structure (LULC, DEM, and slope) alone can reproduce long-term TP patterns, suggesting that watershed structure is a primary control on phosphorus loads and that land-use planning can serve as an effective long-term management tool independent of short-term meteorological variation. Scenario-based simulations further showed that forest-to-residential conversions slightly lowered TP, whereas industrial and agricultural conversions—particularly to paddy fields—substantially increased TP, with central clusters producing the largest impacts. Overall, the CNN–SHAP–IG framework provides local interpretability and domain-wide robustness, representing a methodological advancement over proportion-based studies and offering a practical decision-support tool for watershed management and spatially informed land-use planning to mitigate phosphorus pollution.

本研究提出了一个空间显式框架，通过将卷积神经网络（CNN）与可解释的人工智能技术相结合，预测流中总磷（TP）浓度。利用22个详细的土地利用和土地覆盖（LULC）类别、数字高程模型（DEM）和坡度，对韩国4个主要河流流域608个监测站的5年平均河流TP数据进行CNN-DNN模型训练。该模型优于仅比例回归，突出了空间LULC表示的附加价值。与以往主要依赖于比例LULC指标的TP预测研究不同，我们的方法通过基于cnn的特征提取保留了LULC斑块的空间排列，并集成了SHapley加性解释（SHAP）和集成梯度（IG），以提供斑块级影响的空间明确归因。这使得检测近流集群效应和景观配置，不能被比例模型捕获。基于SHAP的模型解释显示，居住区总体上降低了TP，而商业和公用事业区在靠近溪流时增加了TP。在坡度和水文连通性的调节下，水田对农业流域的正向影响始终是最强的。森林在很大程度上缓冲了全磷，尽管混合林分和邻近的工业地块偶尔会增加养分负荷。为了增强稳健性，IG分析与三个基线显示一致的归因模式，加强了SHAP的发现。研究结果表明，空间流域结构（LULC、DEM和坡度）单独能够重现长期TP格局，表明流域结构是磷负荷的主要控制因素，土地利用规划可以作为一种有效的长期管理工具，不受短期气象变化的影响。基于场景的模拟进一步表明，森林到住宅的转换略微降低了总磷，而工农业转换（尤其是水田）大幅增加了总磷，其中中心集群产生的影响最大。总体而言，CNN-SHAP-IG框架提供了局部可解释性和全域稳健性，代表了基于比例的研究方法的进步，并为流域管理和空间知情的土地利用规划提供了实用的决策支持工具，以减轻磷污染。

{"title":"Assessing the impact of land use and land cover on predicting in-stream total phosphorus using GIS and machine learning models","authors":"Hye Won Lee , Min Kim , Baehyun Min , Jung Hyun Choi","doi":"10.1016/j.ecoinf.2026.103598","DOIUrl":"10.1016/j.ecoinf.2026.103598","url":null,"abstract":"<div><div>This study presents a spatially explicit framework for predicting in-stream total phosphorus (TP) concentrations by combining a convolutional neural network (CNN) with explainable AI techniques. A CNN–dense neural network (CNN–DNN) model was trained on five-year average in-stream TP data from 608 monitoring stations across the four major river basins of the Republic of Korea, using 22 detailed land use and land cover (LULC) classes, a digital elevation model (DEM), and slope. The model outperformed proportion-only regression, highlighting the added value of spatial LULC representation. Unlike previous TP prediction studies that relied primarily on proportional LULC metrics, our approach preserves the spatial arrangement of LULC patches through CNN-based feature extraction and integrates SHapley Additive exPlanations (SHAP) and Integrated Gradients (IG) to provide spatially explicit attribution of patch-level influences. This enables detection of near-stream clustering effects and landscape configurations that cannot be captured by proportion-only models. Model interpretation with SHAP revealed that residential zones generally reduced TP, while commercial and public utility areas increased it when clustered near streams. Paddy fields consistently exerted the strongest positive influence in agricultural watersheds, modulated by slope and hydrologic connectivity. Forests largely buffered TP, though mixed stands and adjacent industrial parcels occasionally increased nutrient loads. To enhance robustness, IG analysis with three baselines revealed consistent attribution patterns that reinforced SHAP findings. Our results reveal that spatial watershed structure (LULC, DEM, and slope) alone can reproduce long-term TP patterns, suggesting that watershed structure is a primary control on phosphorus loads and that land-use planning can serve as an effective long-term management tool independent of short-term meteorological variation. Scenario-based simulations further showed that forest-to-residential conversions slightly lowered TP, whereas industrial and agricultural conversions—particularly to paddy fields—substantially increased TP, with central clusters producing the largest impacts. Overall, the CNN–SHAP–IG framework provides local interpretability and domain-wide robustness, representing a methodological advancement over proportion-based studies and offering a practical decision-support tool for watershed management and spatially informed land-use planning to mitigate phosphorus pollution.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103598"},"PeriodicalIF":7.3,"publicationDate":"2026-01-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145976635","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

AviFluMap: An interactive tool to assess H5N1 avian influenza incursion risk in Australia via migratory birds AviFluMap：评估H5N1禽流感通过候鸟在澳大利亚入侵风险的互动工具

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-06 DOI: 10.1016/j.ecoinf.2026.103603

Tobias A. Ross , Sara Ryding , Simeon Lisovski , Joris Driessen , Emily Mowat , Stephanie Todd , Chris Purnell , Aaron Spence , Simone Vitali , Hui Yu , Marcel Klaassen

The current panzootic of clade 2.3.4.4b H5Nx high pathogenicity avian influenza (H5 HPAI bird flu) has resulted in unprecedented global impacts on both wild bird populations and poultry industries. Despite the virus' near-global circulation, Australia remains free of this strain. In response to the need for proactive biosecurity and conservation planning, we developed AviFluMap, an interactive tool that integrates global data on H5 HPAI events in birds, migratory bird pathways, species susceptibility assessments, and bird aggregation maps to evaluate the incursion risk and establishment of H5 bird flu via wild birds, with special reference to Australia. AviFluMap (https://hpairisk.deakin.edu.au) provides a transparent, data-driven platform for use by a range of stake holders such as wildlife managers, government agencies, researchers, and livestock industry, to support H5 bird flu preparedness and response planning. This article outlines the structure and functionality of AviFluMap, its data sources and methodology, and its role in informing risk-based surveillance and preparedness strategies.

目前发生的2.3.4.4b支H5Nx高致病性禽流感（H5 HPAI禽流感）的大流行，对野生鸟类种群和家禽业造成了前所未有的全球影响。尽管这种病毒几乎在全球范围内传播，但澳大利亚仍然没有这种病毒。为了响应主动生物安全和保护规划的需要，我们开发了AviFluMap，这是一个交互式工具，集成了全球鸟类H5高致病性禽流感事件、候鸟途径、物种易感性评估和鸟类聚集图的数据，以评估野生鸟类入侵风险和H5禽流感的建立，并特别参考了澳大利亚。AviFluMap （https://hpairisk.deakin.edu.au）提供了一个透明的数据驱动平台，供野生动物管理者、政府机构、研究人员和畜牧业等一系列利益攸关方使用，以支持H5禽流感的防范和应对规划。本文概述了AviFluMap的结构和功能、数据来源和方法，以及它在告知基于风险的监测和准备策略方面的作用。

{"title":"AviFluMap: An interactive tool to assess H5N1 avian influenza incursion risk in Australia via migratory birds","authors":"Tobias A. Ross , Sara Ryding , Simeon Lisovski , Joris Driessen , Emily Mowat , Stephanie Todd , Chris Purnell , Aaron Spence , Simone Vitali , Hui Yu , Marcel Klaassen","doi":"10.1016/j.ecoinf.2026.103603","DOIUrl":"10.1016/j.ecoinf.2026.103603","url":null,"abstract":"<div><div>The current panzootic of clade 2.3.4.4b H5Nx high pathogenicity avian influenza (H5 HPAI bird flu) has resulted in unprecedented global impacts on both wild bird populations and poultry industries. Despite the virus' near-global circulation, Australia remains free of this strain. In response to the need for proactive biosecurity and conservation planning, we developed AviFluMap, an interactive tool that integrates global data on H5 HPAI events in birds, migratory bird pathways, species susceptibility assessments, and bird aggregation maps to evaluate the incursion risk and establishment of H5 bird flu via wild birds, with special reference to Australia. AviFluMap (<span><span>https://hpairisk.deakin.edu.au</span><svg><path></path></svg></span>) provides a transparent, data-driven platform for use by a range of stake holders such as wildlife managers, government agencies, researchers, and livestock industry, to support H5 bird flu preparedness and response planning. This article outlines the structure and functionality of AviFluMap, its data sources and methodology, and its role in informing risk-based surveillance and preparedness strategies.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103603"},"PeriodicalIF":7.3,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145977421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Bird diversity responses to mega-events influence: A case study of the Beijing 2022 Winter Olympics 大型事件影响下的鸟类多样性响应——以2022年北京冬奥会为例

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-06 DOI: 10.1016/j.ecoinf.2025.103594

Mingzhao Yu , Luqin Yin , Kunpeng Yi

Mega-events such as the Olympic Games are catalysts for rapid regional transformation, but their ecological impacts remain underexamined, particularly with respect to biodiversity. Based on over 44,000 citizen science records from 2018 to 2024 across Chongli and Yanqing Districts, this study evaluated the impact of the Beijing 2022 Winter Olympic Games on bird diversity. We quantified temporal changes in bird diversity and evaluated the relative influence of environmental factors before and after the Games. Although there was a transient decline during the active preparation and construction phase (2019–2021), the estimated species richness and Simpson diversity eventually increased by 17.66% and 8.95%, respectively, from 2018 to 2024. Post-Olympic gains in bird abundance were significant for water birds, migratory species, omnivores, and herbivores. Generalized linear models demonstrated that habitat factors, particularly vegetation and water coverage, significantly affected bird diversity. Sensitivity analysis further identified the percentage of tall vegetation as the most influential variable in the richness model (main effect: 0.25; total effect: 0.58). These results establish habitat factors as the primary drivers of bird diversity, not only outweighing the negative effects of human disturbance but also exhibiting substantially greater predictive stability than anthropogenic variables. Moreover, bird communities exhibited increased ecological resilience in the post-Olympic period, with reduced sensitivity to environmental factors. These findings underscore the potential for integrating biodiversity conservation into mega-event planning and offer actionable insights for future sustainable urban and regional development strategies.

像奥运会这样的大型活动是区域快速转型的催化剂，但其生态影响仍未得到充分研究，特别是在生物多样性方面。基于2018年至2024年崇礼和延庆地区4.4万多条公民科学记录，本研究评估了北京2022年冬奥会对鸟类多样性的影响。我们量化了鸟类多样性的时间变化，并评估了奥运会前后环境因素的相对影响。虽然在积极准备和建设阶段（2019-2021年）有短暂的下降，但物种丰富度和Simpson多样性最终在2018 - 2024年分别增加了17.66%和8.95%。奥运会后鸟类数量的增加对水鸟、候鸟、杂食动物和食草动物都很重要。广义线性模型表明，生境因子，特别是植被和水覆盖对鸟类多样性有显著影响。敏感性分析进一步确定高植被百分比是丰富度模型中影响最大的变量（主效应为0.25，总效应为0.58）。这些结果表明，生境因子是鸟类多样性的主要驱动因素，不仅超过了人为干扰的负面影响，而且比人为变量表现出更大的预测稳定性。此外，鸟类群落在后奥运时期表现出增强的生态恢复力，对环境因素的敏感性降低。这些发现强调了将生物多样性保护纳入大型活动规划的潜力，并为未来的可持续城市和区域发展战略提供了可行的见解。

{"title":"Bird diversity responses to mega-events influence: A case study of the Beijing 2022 Winter Olympics","authors":"Mingzhao Yu , Luqin Yin , Kunpeng Yi","doi":"10.1016/j.ecoinf.2025.103594","DOIUrl":"10.1016/j.ecoinf.2025.103594","url":null,"abstract":"<div><div>Mega-events such as the Olympic Games are catalysts for rapid regional transformation, but their ecological impacts remain underexamined, particularly with respect to biodiversity. Based on over 44,000 citizen science records from 2018 to 2024 across Chongli and Yanqing Districts, this study evaluated the impact of the Beijing 2022 Winter Olympic Games on bird diversity. We quantified temporal changes in bird diversity and evaluated the relative influence of environmental factors before and after the Games. Although there was a transient decline during the active preparation and construction phase (2019–2021), the estimated species richness and Simpson diversity eventually increased by 17.66% and 8.95%, respectively, from 2018 to 2024. Post-Olympic gains in bird abundance were significant for water birds, migratory species, omnivores, and herbivores. Generalized linear models demonstrated that habitat factors, particularly vegetation and water coverage, significantly affected bird diversity. Sensitivity analysis further identified the percentage of tall vegetation as the most influential variable in the richness model (main effect: 0.25; total effect: 0.58). These results establish habitat factors as the primary drivers of bird diversity, not only outweighing the negative effects of human disturbance but also exhibiting substantially greater predictive stability than anthropogenic variables. Moreover, bird communities exhibited increased ecological resilience in the post-Olympic period, with reduced sensitivity to environmental factors. These findings underscore the potential for integrating biodiversity conservation into mega-event planning and offer actionable insights for future sustainable urban and regional development strategies.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103594"},"PeriodicalIF":7.3,"publicationDate":"2026-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925326","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Remote sensing-based sea ice concentration estimation via weighted deep learning networks 基于加权深度学习网络的遥感海冰浓度估算

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-05 DOI: 10.1016/j.ecoinf.2026.103597

Marzuraikah Mohd Stofa, Mohd Asyraf Zulkifley

Sea ice concentration is an essential monitoring parameter in climate change processes and in navigation and environmental planning in polar regions. Although conventional encoder–decoder of convolutional neural networks perform well in sea ice segmentation, sharp spatial edges and marginal ice zones are difficult to segment without overburdening the computational overhead. A new architecture named feedforward weighted group convolution (FF-WGC), which combines group convolution, asymmetric channel weighting, and lightweight feedforward augmentation, is developed in this research to improve the performance and efficiency of segmentation. The FF-WGC model performs selective discriminative feature group emphasis and intragroup feature learning by progressively involving convolutional transformations. Tests on Sentinel-2 imagery and Canadian Ice Service charts in Hudson Bay reveal that FF-WGC outperforms the current segmentation baselines DeepLab, SegNet, and FCDenseNet in all major indicator measures, including the intersection over union (IoU), accuracy, precision, recall, and F1 score. The proposed architecture produced an IoU of 47.5 % with non-statistically significant additional complexity and offers a good compromise between the required performance and prolonged operation in large-scale sea ice applications. The current work can be extended by incorporating attention mechanisms and testing over multitemporal sequences to forecast sea ice conditions, with validation in other Arctic and Antarctic regions.

海冰浓度是气候变化过程、极地航行和环境规划的重要监测参数。传统的卷积神经网络编解码器在海冰分割中表现良好，但在不增加计算负担的情况下，难以分割尖锐的空间边缘和边缘冰区。为了提高分割的性能和效率，本研究提出了一种新的前馈加权群卷积（FF-WGC）结构，该结构结合了群卷积、非对称信道加权和轻量级前馈增强。FF-WGC模型通过逐步涉及卷积变换进行选择性判别特征组强调和组内特征学习。在哈德逊湾对Sentinel-2图像和加拿大Ice Service图表进行的测试表明，FF-WGC在所有主要指标指标上都优于当前的分割基线DeepLab、SegNet和FCDenseNet，包括交汇（IoU）、准确性、精密度、查全率和F1分数。所提出的架构产生了47.5%的IoU，具有非统计上显著的额外复杂性，并且在大规模海冰应用中提供了所需性能和长时间操作之间的良好折衷。目前的工作可以通过纳入关注机制和对多时间序列的测试来扩展，以预测海冰状况，并在其他北极和南极地区进行验证。

{"title":"Remote sensing-based sea ice concentration estimation via weighted deep learning networks","authors":"Marzuraikah Mohd Stofa, Mohd Asyraf Zulkifley","doi":"10.1016/j.ecoinf.2026.103597","DOIUrl":"10.1016/j.ecoinf.2026.103597","url":null,"abstract":"<div><div>Sea ice concentration is an essential monitoring parameter in climate change processes and in navigation and environmental planning in polar regions. Although conventional encoder–decoder of convolutional neural networks perform well in sea ice segmentation, sharp spatial edges and marginal ice zones are difficult to segment without overburdening the computational overhead. A new architecture named feedforward weighted group convolution (FF-WGC), which combines group convolution, asymmetric channel weighting, and lightweight feedforward augmentation, is developed in this research to improve the performance and efficiency of segmentation. The FF-WGC model performs selective discriminative feature group emphasis and intragroup feature learning by progressively involving convolutional transformations. Tests on Sentinel-2 imagery and Canadian Ice Service charts in Hudson Bay reveal that FF-WGC outperforms the current segmentation baselines DeepLab, SegNet, and FCDenseNet in all major indicator measures, including the intersection over union (IoU), accuracy, precision, recall, and F1 score. The proposed architecture produced an IoU of 47.5 % with non-statistically significant additional complexity and offers a good compromise between the required performance and prolonged operation in large-scale sea ice applications. The current work can be extended by incorporating attention mechanisms and testing over multitemporal sequences to forecast sea ice conditions, with validation in other Arctic and Antarctic regions.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"94 ","pages":"Article 103597"},"PeriodicalIF":7.3,"publicationDate":"2026-01-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146025992","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Predicting phytoplankton community in a large floodplain lake using interpretable machine learning 利用可解释的机器学习预测大型洪泛区湖泊中的浮游植物群落

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-04 DOI: 10.1016/j.ecoinf.2026.103596

Xi Luo, Haibin Cai, Jingqiao Mao

Floodplain lakes are highly dynamic ecosystems where hydrological fluctuations drive complex phytoplankton responses that are challenging to predict. In this study, a random forest (RF) framework was developed to predict phytoplankton succession in Poyang Lake, the largest freshwater lake in China, using integrated hydrological, hydrodynamic, meteorological, and physicochemical data collected from 2019 to 2021. The RF models exhibited robust predictive performance, with phytoplankton density identified as the most reliable indicator for capturing community variability (R²=0.62–0.75), higher than biomass (R²=0.47–0.72) or Chlorophyll-a (R²=0.70). SHapley Additive exPlanations (SHAP) analysis further revealed that lagged hydrometeorological conditions, specifically 30-day inflow and 15-day cumulative precipitation, were the dominant drivers of total phytoplankton density, whereas suspended solids (SS) and time period primarily influenced community composition. These findings provide novel insights into phytoplankton studies in Poyang Lake, emphasize the key role of hydrological lags in shaping phytoplankton dynamics, and offer theoretical support for eutrophication management in Poyang Lake.

泛滥平原湖泊是高度动态的生态系统，其中水文波动驱动复杂的浮游植物反应，难以预测。利用2019 - 2021年收集的水文、水动力、气象和理化数据，构建随机森林（RF）框架，预测中国最大的淡水湖鄱阳湖浮游植物演替。RF模型具有较强的预测能力，其中浮游植物密度被认为是捕获群落变异的最可靠指标（R2= 0.62-0.75），高于生物量（R2= 0.47-0.72）或叶绿素a （R2=0.70）。SHapley加性解释（SHAP）进一步揭示了滞后的水文气象条件（特别是30 d入流和15 d累积降水）是浮游植物总密度的主要驱动因素，而悬浮物（SS）和时间段主要影响群落组成。这些发现为鄱阳湖浮游植物的研究提供了新的思路，强调了水文滞后在浮游植物动态形成中的关键作用，并为鄱阳湖富营养化管理提供了理论支持。

{"title":"Predicting phytoplankton community in a large floodplain lake using interpretable machine learning","authors":"Xi Luo, Haibin Cai, Jingqiao Mao","doi":"10.1016/j.ecoinf.2026.103596","DOIUrl":"10.1016/j.ecoinf.2026.103596","url":null,"abstract":"<div><div>Floodplain lakes are highly dynamic ecosystems where hydrological fluctuations drive complex phytoplankton responses that are challenging to predict. In this study, a random forest (RF) framework was developed to predict phytoplankton succession in Poyang Lake, the largest freshwater lake in China, using integrated hydrological, hydrodynamic, meteorological, and physicochemical data collected from 2019 to 2021. The RF models exhibited robust predictive performance, with phytoplankton density identified as the most reliable indicator for capturing community variability (R<sup>2</sup>=0.62–0.75), higher than biomass (R<sup>2</sup>=0.47–0.72) or Chlorophyll-a (R<sup>2</sup>=0.70). SHapley Additive exPlanations (SHAP) analysis further revealed that lagged hydrometeorological conditions, specifically 30-day inflow and 15-day cumulative precipitation, were the dominant drivers of total phytoplankton density, whereas suspended solids (SS) and time period primarily influenced community composition. These findings provide novel insights into phytoplankton studies in Poyang Lake, emphasize the key role of hydrological lags in shaping phytoplankton dynamics, and offer theoretical support for eutrophication management in Poyang Lake.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103596"},"PeriodicalIF":7.3,"publicationDate":"2026-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

SALMA: A machine learning tool for precise leaf morphology measurements SALMA：用于精确测量叶片形态的机器学习工具

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2026-01-02 DOI: 10.1016/j.ecoinf.2025.103592

Ilya Shabanov , Julie R Deslippe , Andrew Lensen

Leaf area is a critical plant functional trait, widely used for understanding plant responses to climate change, ecosystem productivity, and species' adaptive strategies. Inaccurate leaf area measurements compromise the accuracy of other traits normalised by area, such as foliar chemical traits, respiration, and photosynthesis. However, existing measurement methods are ineffective for small-leaved plants and often necessitate manual processing, which limits sample sizes and potentially obscures subtle trait-environment relationships. We developed SALMA (Semi-Automated Leaf Morphological Analysis), which employs logistic regression trained on one to four human-generated examples per species to delineate leaf boundaries for that species accurately. SALMA's training step adapts to species-specific features by integrating multiple characteristics, such as colour variations and edge details. The approach is validated on an extensive dataset (64 species, 3332 images) that covers 91.4 % of the worldwide leaf area variation, as well as two smaller datasets comprising low-quality photographs of morphologically complex or damaged leaves. SALMA consistently achieved leaf area errors 2 to 15 times lower than existing algorithms and a theoretical upper bound of any grayscale intensity-based method. Critically, we identify a previously overlooked power-law relationship between leaf area and measurement error, demonstrating that existing methods may overestimate leaf area by at least 5 % for 43 % of global species, whereas SALMA achieves comparable errors for only 2.1 % of species. SALMA is a standalone software with an intuitive interface that supports parallel processing, making it accessible for large-scale ecological studies globally. It can potentially enhance the quality of trait datasets and facilitate large-scale sampling, thereby advancing our understanding of plant-environment interactions. Our published dataset contains manually created binary segmentations of leaves and background, providing a baseline for future leaf measurement algorithms.

叶面积是一项重要的植物功能性状，广泛用于了解植物对气候变化的响应、生态系统生产力和物种适应策略。不准确的叶面积测量会损害其他性状的准确性，如叶化学性状、呼吸作用和光合作用。然而，现有的测量方法对小叶植物是无效的，通常需要人工处理，这限制了样本量，并可能模糊微妙的性状-环境关系。我们开发了SALMA（半自动叶片形态分析），它使用逻辑回归训练每个物种的一到四个人类生成的例子来准确地描绘该物种的叶片边界。SALMA的训练步骤通过整合多种特征（如颜色变化和边缘细节）来适应物种特定的特征。该方法在一个广泛的数据集（64种，3332张图像）上进行了验证，该数据集覆盖了全球91.4%的叶面积变化，以及两个较小的数据集，其中包括形态复杂或受损叶片的低质量照片。SALMA始终实现叶面积误差比现有算法低2到15倍，并且是任何基于灰度强度的方法的理论上限。重要的是，我们确定了以前被忽视的叶面积与测量误差之间的幂律关系，表明现有方法可能对全球43%的物种高估了至少5%的叶面积，而SALMA仅对2.1%的物种实现了类似的误差。SALMA是一个独立的软件，具有直观的界面，支持并行处理，使其可用于全球大规模的生态研究。它可以潜在地提高性状数据集的质量，促进大规模采样，从而提高我们对植物-环境相互作用的理解。我们发布的数据集包含手动创建的叶片和背景的二值分割，为未来的叶片测量算法提供基线。

{"title":"SALMA: A machine learning tool for precise leaf morphology measurements","authors":"Ilya Shabanov , Julie R Deslippe , Andrew Lensen","doi":"10.1016/j.ecoinf.2025.103592","DOIUrl":"10.1016/j.ecoinf.2025.103592","url":null,"abstract":"<div><div>Leaf area is a critical plant functional trait, widely used for understanding plant responses to climate change, ecosystem productivity, and species' adaptive strategies. Inaccurate leaf area measurements compromise the accuracy of other traits normalised by area, such as foliar chemical traits, respiration, and photosynthesis. However, existing measurement methods are ineffective for small-leaved plants and often necessitate manual processing, which limits sample sizes and potentially obscures subtle trait-environment relationships. We developed SALMA (Semi-Automated Leaf Morphological Analysis), which employs logistic regression trained on one to four human-generated examples per species to delineate leaf boundaries for that species accurately. SALMA's training step adapts to species-specific features by integrating multiple characteristics, such as colour variations and edge details. The approach is validated on an extensive dataset (64 species, 3332 images) that covers 91.4 % of the worldwide leaf area variation, as well as two smaller datasets comprising low-quality photographs of morphologically complex or damaged leaves. SALMA consistently achieved leaf area errors 2 to 15 times lower than existing algorithms and a theoretical upper bound of any grayscale intensity-based method. Critically, we identify a previously overlooked power-law relationship between leaf area and measurement error, demonstrating that existing methods may overestimate leaf area by at least 5 % for 43 % of global species, whereas SALMA achieves comparable errors for only 2.1 % of species. SALMA is a standalone software with an intuitive interface that supports parallel processing, making it accessible for large-scale ecological studies globally. It can potentially enhance the quality of trait datasets and facilitate large-scale sampling, thereby advancing our understanding of plant-environment interactions. Our published dataset contains manually created binary segmentations of leaves and background, providing a baseline for future leaf measurement algorithms.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103592"},"PeriodicalIF":7.3,"publicationDate":"2026-01-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925403","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

From Leaves to Breezes: Machine learning based prediction of nitrogen dioxide concentration from surrounding urban greenery and meteorological, spatial, and traffic characteristics in Berlin, Germany 从树叶到微风：基于机器学习的二氧化氮浓度预测，从周围的城市绿化和气象、空间和交通特征，在德国柏林

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-30 DOI: 10.1016/j.ecoinf.2025.103568

Richard Schmidt , L.L. Sharon Ong

This study compares two machine learning models, a Random Forest (RF) and a spatial Graph Neural Network (GNN), for predicting nitrogen dioxide (NO

_{2}

) concentrations across diverse urban conditions in Berlin, Germany. Therefore, both models use information on local land-use characteristics, meteorological conditions, and seasonal greenery, which enables a post-hoc analysis of high-concentration scenarios under varying environmental factors. Unlike most previous approaches to air-pollution estimation, this study explicitly considers the interaction between urban greenery and its seasonal variation. The analysis is based on a self-curated, high-resolution site-level environmental dataset that captures hourly NO

_{2}

observations from sixteen monitoring stations across Berlin in 2023 with detailed land-use, traffic, and architectural data obtained from the Berlin Geoportal. This dataset is supplemented with multiple meteorological records from the Deutscher Wetterdienst (DWD). While both models achieve comparable accuracy (R

^{2}

\approx

0.6), the GNN shows a tendency toward less variation of predictive accuracy across test sites, suggesting potential spatial robustness. For explainability, only the RF model allows for local interpretability via Shapley values, which indicate that urban greenery helps mitigate NO

_{2}

levels depending on seasonal changes in leaf area. However, additional statistical testing does not support this observed trend. Beyond the conducted assessment, this research contributes a comprehensive environmental dataset that links air quality, land-use, and meteorological variables at hourly resolution. This resource supports future investigations into how environmental and spatial factors jointly influence pollutant dispersion and decomposition in urban environments.

本研究比较了两种机器学习模型，随机森林（RF）和空间图神经网络（GNN），用于预测德国柏林不同城市条件下的二氧化氮（NO2）浓度。因此，这两种模式都使用了当地土地利用特征、气象条件和季节性绿化的信息，从而能够对不同环境因素下的高浓度情景进行事后分析。与以往大多数空气污染估算方法不同，本研究明确考虑了城市绿化与其季节变化之间的相互作用。该分析基于自定义的高分辨率站点级环境数据集，该数据集捕获了2023年柏林16个监测站的每小时二氧化氮观测数据，并从柏林地质门户获得了详细的土地利用、交通和建筑数据。该数据集补充了来自德国湿地（DWD）的多个气象记录。虽然两种模型的精度相当（R2≈0.6），但GNN在不同测试地点的预测精度变化趋势较小，表明潜在的空间稳健性。就可解释性而言，只有RF模型允许通过Shapley值进行局部可解释性，该值表明城市绿化有助于根据叶面积的季节变化降低NO2水平。然而，额外的统计测试并不支持这种观察到的趋势。除了进行评估之外，本研究还提供了一个综合的环境数据集，该数据集以小时分辨率将空气质量、土地利用和气象变量联系起来。这一资源支持未来研究环境和空间因素如何共同影响污染物在城市环境中的扩散和分解。

{"title":"From Leaves to Breezes: Machine learning based prediction of nitrogen dioxide concentration from surrounding urban greenery and meteorological, spatial, and traffic characteristics in Berlin, Germany","authors":"Richard Schmidt , L.L. Sharon Ong","doi":"10.1016/j.ecoinf.2025.103568","DOIUrl":"10.1016/j.ecoinf.2025.103568","url":null,"abstract":"<div><div>This study compares two machine learning models, a Random Forest (RF) and a spatial Graph Neural Network (GNN), for predicting nitrogen dioxide (NO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span>) concentrations across diverse urban conditions in Berlin, Germany. Therefore, both models use information on local land-use characteristics, meteorological conditions, and seasonal greenery, which enables a post-hoc analysis of high-concentration scenarios under varying environmental factors. Unlike most previous approaches to air-pollution estimation, this study explicitly considers the interaction between urban greenery and its seasonal variation. The analysis is based on a self-curated, high-resolution site-level environmental dataset that captures hourly NO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span> observations from sixteen monitoring stations across Berlin in 2023 with detailed land-use, traffic, and architectural data obtained from the <span><span>Berlin Geoportal</span><svg><path></path></svg></span>. This dataset is supplemented with multiple meteorological records from the <span><span>Deutscher Wetterdienst (DWD)</span><svg><path></path></svg></span>. While both models achieve comparable accuracy (R<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span> <span><math><mo>≈</mo></math></span> 0.6), the GNN shows a tendency toward less variation of predictive accuracy across test sites, suggesting potential spatial robustness. For explainability, only the RF model allows for local interpretability via Shapley values, which indicate that urban greenery helps mitigate NO<span><math><msub><mrow></mrow><mrow><mn>2</mn></mrow></msub></math></span> levels depending on seasonal changes in leaf area. However, additional statistical testing does not support this observed trend. Beyond the conducted assessment, this research contributes a comprehensive environmental dataset that links air quality, land-use, and meteorological variables at hourly resolution. This resource supports future investigations into how environmental and spatial factors jointly influence pollutant dispersion and decomposition in urban environments.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103568"},"PeriodicalIF":7.3,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925401","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Deconstructing niche shifts: Using native and non-native records to assess the invasion potential of geographically constrained species 解构生态位转移：利用本地和非本地记录评估地理受限物种的入侵潜力

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-30 DOI: 10.1016/j.ecoinf.2025.103591

Melinda S. Trudgen , John K. Scott , Hans Lambers , Bruce L. Webber

Geographically-constrained species typically occupy small climatic niches in their native range, and are therefore often assumed to pose a low risk of becoming invasive. However, the association between range size and climatic niche is correlative, and the cause of the constraint, whether it is physiological limits, competition, or dispersal limitations, impacts their risk of becoming invasive. We investigated the interplay between geographic range and niche breadth for the South American tree species Tipuana tipu (Fabaceae), and quantified how the spatial projection of the realised niche may be impacted by this relationship. We constructed a global dataset of location records, including datasets of both non-native and cultivated plants. We constructed a correlative species distribution model (CLIMEX Match Climates), and used ExDet and Ecospat to investigate and characterise niche shift. We found that T. tipu had a large climatic niche, despite having a small geographic native range. Projecting the correlative model indicated a large proportion of Australia and the globe were climatically suitable for T. tipu. Moderate niche shift occurred in all introduced regions, with a significantly greater niche shift amongst datasets of cultivated plants. Geographically-constrained species may have large climatic niches, thereby posing an unanticipated invasion risk in their non-native range. Correlative modelling is a useful tool for modelling species distribution; however, a small geographic range does not guarantee a small climatic niche. Niche shifting may also affect correlative modelling results and their interpretation.

受地理限制的物种通常在其原生范围内占据较小的气候生态位，因此通常被认为具有较低的入侵风险。然而，范围大小与气候生态位之间的关系是相关的，并且约束的原因，无论是生理限制，竞争还是分散限制，都会影响它们成为入侵的风险。本文研究了南美树种Tipuana tipu （Fabaceae）的地理范围和生态位宽度之间的相互作用，并量化了这种关系如何影响实现生态位的空间投影。我们构建了一个全球位置记录数据集，包括非本地和栽培植物的数据集。我们构建了CLIMEX Match Climates相关物种分布模型，并利用ExDet和ecospit对生态位变化进行了研究和表征。我们发现T. tipu有一个很大的气候生态位，尽管它的地理原生范围很小。对相关模式的预测表明，澳大利亚和全球大部分地区的气候都适合T. tipu生长。所有引入区均发生中度生态位转移，其中栽培植物数据集的生态位转移明显更大。受地理限制的物种可能具有较大的气候生态位，从而在其非本地范围内造成意想不到的入侵风险。相关模型是模拟物种分布的有效工具；然而，一个小的地理范围并不保证一个小的气候位。生态位移动也可能影响相关的模拟结果及其解释。

{"title":"Deconstructing niche shifts: Using native and non-native records to assess the invasion potential of geographically constrained species","authors":"Melinda S. Trudgen , John K. Scott , Hans Lambers , Bruce L. Webber","doi":"10.1016/j.ecoinf.2025.103591","DOIUrl":"10.1016/j.ecoinf.2025.103591","url":null,"abstract":"<div><div>Geographically-constrained species typically occupy small climatic niches in their native range, and are therefore often assumed to pose a low risk of becoming invasive. However, the association between range size and climatic niche is correlative, and the cause of the constraint, whether it is physiological limits, competition, or dispersal limitations, impacts their risk of becoming invasive. We investigated the interplay between geographic range and niche breadth for the South American tree species <em>Tipuana tipu</em> (Fabaceae), and quantified how the spatial projection of the realised niche may be impacted by this relationship. We constructed a global dataset of location records, including datasets of both non-native and cultivated plants. We constructed a correlative species distribution model (CLIMEX Match Climates), and used ExDet and Ecospat to investigate and characterise niche shift. We found that <em>T. tipu</em> had a large climatic niche, despite having a small geographic native range. Projecting the correlative model indicated a large proportion of Australia and the globe were climatically suitable for <em>T. tipu</em>. Moderate niche shift occurred in all introduced regions, with a significantly greater niche shift amongst datasets of cultivated plants. Geographically-constrained species may have large climatic niches, thereby posing an unanticipated invasion risk in their non-native range. Correlative modelling is a useful tool for modelling species distribution; however, a small geographic range does not guarantee a small climatic niche. Niche shifting may also affect correlative modelling results and their interpretation.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103591"},"PeriodicalIF":7.3,"publicationDate":"2025-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925402","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A lightweight Mamba-frequency fusion algorithm for underwater image enhancement under physical model constraints 物理模型约束下水下图像增强的轻量mamba频率融合算法

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-29 DOI: 10.1016/j.ecoinf.2025.103590

Hui Zhou , Meiwei Kong , Ruizhi Wang , Qunhui Yang

Underwater optical images are crucial data for ecological environment monitoring and marine biology research. However, due to the absorption and scattering of light by water, the image quality often suffers severe degradation, directly affecting the accuracy of species identification and habitat monitoring in ecological studies. The existing underwater image enhancement (UIE) methods have two major problems: non-physical methods are prone to producing unreasonable enhancements and overfitting, while physical model methods have high computational costs, making them difficult to deploy on resource-constrained underwater devices. Therefore, there is an urgent need to design a lightweight model for underwater environments that efficiently balances color correction and detail preservation. We propose a lightweight physical model-constrained UIE network, namely the lightweight Mamba-frequency fusion (LMF) algorithm. It utilizes an underwater optical imaging model to ensure the physical rationality of the imaging process. The core innovation lies in its lightweight dual-path architecture. It uses the linear complexity Mamba module to efficiently model global features, combines convolution to extract local details, and adopts a frequency-domain fusion strategy to fuse global, local, and channel features in the frequency domain, thereby alleviating detail loss caused by scattering and color distortion caused by absorption at a low computational cost. Experiments conducted on two public datasets show that LMF outperforms mainstream methods in image quality assessment metrics, while significantly reducing the number of parameters and computational complexity. This advancement provides a technically feasible approach for high-fidelity visual data acquisition in ecological monitoring applications.

水下光学图像是生态环境监测和海洋生物学研究的重要数据。然而，由于水体对光的吸收和散射，图像质量往往会严重下降，直接影响生态研究中物种鉴定和生境监测的准确性。现有的水下图像增强（UIE）方法存在两个主要问题：非物理方法容易产生不合理的增强和过拟合，而物理模型方法计算成本高，难以在资源受限的水下设备上部署。因此，迫切需要设计一种用于水下环境的轻型模型，有效地平衡色彩校正和细节保存。我们提出了一种轻量级的物理模型约束的UIE网络，即轻量级的曼巴频率融合（LMF）算法。它采用水下光学成像模型，保证了成像过程的物理合理性。核心创新在于其轻量级双路径架构。利用线性复杂度Mamba模块高效建模全局特征，结合卷积提取局部细节，并采用频域融合策略在频域融合全局、局部和信道特征，以较低的计算成本减轻散射和吸收引起的颜色失真带来的细节损失。在两个公开数据集上进行的实验表明，LMF在图像质量评估指标方面优于主流方法，同时显著减少了参数数量和计算复杂度。这一进展为生态监测应用中的高保真视觉数据采集提供了技术上可行的途径。

{"title":"A lightweight Mamba-frequency fusion algorithm for underwater image enhancement under physical model constraints","authors":"Hui Zhou , Meiwei Kong , Ruizhi Wang , Qunhui Yang","doi":"10.1016/j.ecoinf.2025.103590","DOIUrl":"10.1016/j.ecoinf.2025.103590","url":null,"abstract":"<div><div>Underwater optical images are crucial data for ecological environment monitoring and marine biology research. However, due to the absorption and scattering of light by water, the image quality often suffers severe degradation, directly affecting the accuracy of species identification and habitat monitoring in ecological studies. The existing underwater image enhancement (UIE) methods have two major problems: non-physical methods are prone to producing unreasonable enhancements and overfitting, while physical model methods have high computational costs, making them difficult to deploy on resource-constrained underwater devices. Therefore, there is an urgent need to design a lightweight model for underwater environments that efficiently balances color correction and detail preservation. We propose a lightweight physical model-constrained UIE network, namely the lightweight Mamba-frequency fusion (LMF) algorithm. It utilizes an underwater optical imaging model to ensure the physical rationality of the imaging process. The core innovation lies in its lightweight dual-path architecture. It uses the linear complexity Mamba module to efficiently model global features, combines convolution to extract local details, and adopts a frequency-domain fusion strategy to fuse global, local, and channel features in the frequency domain, thereby alleviating detail loss caused by scattering and color distortion caused by absorption at a low computational cost. Experiments conducted on two public datasets show that LMF outperforms mainstream methods in image quality assessment metrics, while significantly reducing the number of parameters and computational complexity. This advancement provides a technically feasible approach for high-fidelity visual data acquisition in ecological monitoring applications.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103590"},"PeriodicalIF":7.3,"publicationDate":"2025-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925263","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Diagnosis of pine wilt disease using unattended UAV hyperspectral imager: A comparison of discrete bands and continuous spectrum -based methods 无人值守无人机高光谱成像仪诊断松材萎蔫病：基于离散波段和连续光谱方法的比较

IF 7.3 2区环境科学与生态学 Q1 ECOLOGY

Ecological Informatics

Pub Date : 2025-12-29 DOI: 10.1016/j.ecoinf.2025.103589

Minghao Sui , Xing Wang , Shuo Yang , Linyuan Li , Wei Li , Chang Nie , Huaguo Huang

Unattended UAV systems offer an effective solution for high-frequency and high-resolution monitoring of forest disease with minimal human intervention. Deploying such systems, however, requires classification methods that achieve high accuracy while remaining robust to spectral interference and computationally efficient for onboard or edge processing. Balancing these requirements is critical for practical pine wilt disease (PWD) surveillance, yet has received limited attention. In this study, we systematically evaluated four spectral analysis approaches for PWD detection using hyperspectral imagery acquired from an unattended UAV platform. These methods include: (1) a newly developed vegetation index, PWDAI, based on discrete spectral bands; (2) a Spectral Angle Mapper (SAM) method utilizing a band-optimized, partially continuous spectrum; (3) a Partial Least Squares Discriminant Analysis (PLS–DA) model and (4) a one-dimensional convolutional neural network (1D-CNN) model, both leveraging the full hyperspectral spectrum (400-1000 nm). Using object-based samples of healthy, diseased, shadow, and mixed classes, we identified PWD-sensitive spectral features in both discrete and continuous forms, which were subsequently used for the design of PWDAI and application of SAM method. Comparison results show that 1D-CNN, utilizing the full spectrum, achieved the highest classification accuracy (overall accuracy: 92.34 %, Kappa: 0.85) and exhibited strong resilience to shadow and mixed pixel interference. PWDAI, by contrast, used only three narrow bands yet achieved competitive accuracy (OA: 83.86 %) with minimal computational cost, offering a practical trade-off for onboard application. Intermediate methods such as SAM and PLS-DA demonstrated moderate performance. These findings suggest that full-spectrum deep learning models are optimal for offline high-precision mapping, while targeted vegetation indices like PWDAI provide interpretable and lightweight alternatives for real-time UAV-based disease screening. The integration of such adaptable spectral classifiers with unattended UAV systems offers a promising pathway for scalable, automated forest health monitoring.

无人值守无人机系统为森林疾病的高频和高分辨率监测提供了有效的解决方案，而人工干预最少。然而，部署这样的系统需要实现高精度的分类方法，同时保持对频谱干扰的鲁棒性，以及机载或边缘处理的计算效率。平衡这些要求对于实际的松树枯萎病（PWD）监测至关重要，但受到的关注有限。在这项研究中，我们系统地评估了四种用于PWD检测的光谱分析方法，这些方法使用了从无人值机无人机平台获取的高光谱图像。这些方法包括：(1)基于离散光谱波段的植被指数PWDAI；(2)利用波段优化、部分连续光谱的光谱角映射（SAM）方法；(3)偏最小二乘判别分析（PLS-DA）模型和(4)一维卷积神经网络（1D-CNN）模型，两者都利用了全高光谱（400-1000 nm）。利用健康、患病、阴影和混合类别的基于对象的样本，我们确定了离散和连续形式的pwd敏感光谱特征，随后将其用于PWDAI的设计和SAM方法的应用。对比结果表明，利用全光谱的1D-CNN分类准确率最高（总体准确率为92.34%，Kappa为0.85），并且对阴影和混合像元干扰具有较强的恢复能力。相比之下，PWDAI仅使用三个窄带，但以最小的计算成本获得了具有竞争力的精度（OA: 83.86%），为机载应用提供了实际的权衡。中间方法如SAM和PLS-DA表现出中等的性能。这些发现表明，全谱深度学习模型是离线高精度制图的最佳选择，而像PWDAI这样的目标植被指数为基于无人机的实时疾病筛查提供了可解释和轻量级的替代方案。这种自适应光谱分类器与无人值守无人机系统的集成为可扩展的自动化森林健康监测提供了一条有前途的途径。

{"title":"Diagnosis of pine wilt disease using unattended UAV hyperspectral imager: A comparison of discrete bands and continuous spectrum -based methods","authors":"Minghao Sui , Xing Wang , Shuo Yang , Linyuan Li , Wei Li , Chang Nie , Huaguo Huang","doi":"10.1016/j.ecoinf.2025.103589","DOIUrl":"10.1016/j.ecoinf.2025.103589","url":null,"abstract":"<div><div>Unattended UAV systems offer an effective solution for high-frequency and high-resolution monitoring of forest disease with minimal human intervention. Deploying such systems, however, requires classification methods that achieve high accuracy while remaining robust to spectral interference and computationally efficient for onboard or edge processing. Balancing these requirements is critical for practical pine wilt disease (PWD) surveillance, yet has received limited attention. In this study, we systematically evaluated four spectral analysis approaches for PWD detection using hyperspectral imagery acquired from an unattended UAV platform. These methods include: (1) a newly developed vegetation index, PWDAI, based on discrete spectral bands; (2) a Spectral Angle Mapper (SAM) method utilizing a band-optimized, partially continuous spectrum; (3) a Partial Least Squares Discriminant Analysis (PLS–DA) model and (4) a one-dimensional convolutional neural network (1D-CNN) model, both leveraging the full hyperspectral spectrum (400-1000 nm). Using object-based samples of healthy, diseased, shadow, and mixed classes, we identified PWD-sensitive spectral features in both discrete and continuous forms, which were subsequently used for the design of PWDAI and application of SAM method. Comparison results show that 1D-CNN, utilizing the full spectrum, achieved the highest classification accuracy (overall accuracy: 92.34 %, Kappa: 0.85) and exhibited strong resilience to shadow and mixed pixel interference. PWDAI, by contrast, used only three narrow bands yet achieved competitive accuracy (OA: 83.86 %) with minimal computational cost, offering a practical trade-off for onboard application. Intermediate methods such as SAM and PLS-DA demonstrated moderate performance. These findings suggest that full-spectrum deep learning models are optimal for offline high-precision mapping, while targeted vegetation indices like PWDAI provide interpretable and lightweight alternatives for real-time UAV-based disease screening. The integration of such adaptable spectral classifiers with unattended UAV systems offers a promising pathway for scalable, automated forest health monitoring.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"93 ","pages":"Article 103589"},"PeriodicalIF":7.3,"publicationDate":"2025-12-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145925261","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0