Ecological Informatics最新文献

{"title":"Safe AI for coral reefs: Benchmarking out-of-distribution detection algorithms for coral reef image surveys","authors":"Mathew Wyatt ,&nbsp;Sharyn Hickey ,&nbsp;Ben Radford ,&nbsp;Manuel Gonzalez-Rivero ,&nbsp;Nader Boutros ,&nbsp;Nikolaus Callow ,&nbsp;Nicole Ryan ,&nbsp;Arjun Chennu ,&nbsp;Mohammed Bennamoun ,&nbsp;James Gilmour","doi":"10.1016/j.ecoinf.2025.103207","DOIUrl":"10.1016/j.ecoinf.2025.103207","url":null,"abstract":"<div><div>Although deep learning has demonstrated significant advances in qualitative domains, deep learning algorithms remain poor at quantifying the uncertainty of their predictions. This is especially true when applied in domains where there is data shift outside of which an algorithm has been trained. This has major implications for the use of deep learning in accurately estimating change in environmental monitoring applications. In the case of image classification for coral reef habitats, time series imagery is rarely consistent due to changing environmental conditions, differing sensors and inconsistencies in data capture, which means traditional machine learning metrics simply do not work when applied to new out of distribution datasets.<ul><li><span>1.</span><span><div>For this reason, we benchmark the latest state-of-the-art OOD (Out Of Distribution) detection algorithms on publicly available coral reef image datasets, and evaluate histogram intersection of outlier scores as an indicator for human intervention.</div></span></li><li><span>2.</span><span><div>We show with a comparative analysis that the performance of OOD detection algorithms is variable, and highly dependent on in-distribution and out-of-distribution data composition. We show that KNN (K-Nearest Neighbour) distance was the most consistent across datasets, followed by Virtual-logit matching (ViM).</div></span></li><li><span>3.</span><span><div>This research shows a compelling example of how a handful of openly available algorithms can easily be used as a complimentary indicator alongside confidence (Softmax probability), in turn providing more efficient and safe deployment of deep learning for rapid reporting of coral reef habitats.</div></span></li></ul></div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"90 ","pages":"Article 103207"},"PeriodicalIF":5.8,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124710","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}

{"title":"Spatio-temporal patterns of holoplanktonic mollusc assemblages and indicator species of hydrodynamic conditions in the northwestern South China Sea","authors":"Junce Liang ,&nbsp;Yuan Liu ,&nbsp;Kaizhi Li ,&nbsp;Yehui Tan ,&nbsp;Jiaxing Liu ,&nbsp;Pengli Xiong ,&nbsp;Yu Zhong","doi":"10.1016/j.ecoinf.2025.103206","DOIUrl":"10.1016/j.ecoinf.2025.103206","url":null,"abstract":"<div><div>Holoplanktonic molluscs are a group of overlooked but ecologically important zooplankton in marine ecosystems that play a crucial role in the marine food web and carbon cycle. However, changes in the marine environment can lead to an increase in the abundance of pelagic molluscs, causing ecological disasters. To understand the environmental factors that structure holoplanktonic mollusc assemblages, their diversity and abundance were analyzed in the northwestern South China Sea (NWSCS). A total of 39 holoplanktonic molluscs were identified, including 24 pteropod and 15 heteropod species. Significant seasonal and regional differences were observed in species diversity and abundance. Species richness was higher in offshore waters than in nearshore waters during summer and winter, whereas species abundance was significantly higher in summer than in winter. High species abundance was mainly concentrated in waters influenced by cyclonic eddies and coastal upwelling during summer. Temperature, salinity and chlorophyll <em>a</em> are important factors for structuring holoplanktonic mollusc assemblages. <em>Creseis acicula</em> was an effective indicator species in nearshore waters in both seasons, whereas <em>Limacina bulimoides</em> and <em>Heliconoides inflatus</em> were the best indicators in offshore waters in summer and winter, respectively. These findings can form a baseline for understanding the distribution of holoplanktonic molluscs in the marginal sea of the northwestern Pacific Ocean and provide a solid foundation for monitoring zooplankton in a changing ocean and for sustainable ecosystem management.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"90 ","pages":"Article 103206"},"PeriodicalIF":5.8,"publicationDate":"2025-05-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144124805","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}

{"title":"A novel method for monitoring Cladophora blooms in Qinghai Lake based on UAV imagery","authors":"Juan Zhang ,&nbsp;Xiaojun Yao ,&nbsp;Xinde Chu ,&nbsp;Hongyu Duan ,&nbsp;Yuxuan Zhang ,&nbsp;Hui Chang ,&nbsp;Ju Huang","doi":"10.1016/j.ecoinf.2025.103210","DOIUrl":"10.1016/j.ecoinf.2025.103210","url":null,"abstract":"<div><div>The outbreak of <em>Cladophora</em> in the nearshore region of Qinghai Lake has significantly affected the surrounding ecological landscape and tourism industry. However, existing satellite remote sensing methods for monitoring <em>Cladophora</em> blooms have several limitations owing to cloud cover in optical images and the small areas of <em>Cladophora</em> bloom patches, making it challenging to obtain detailed dynamic characteristics of these blooms. Unmanned aerial vehicles (UAVs) equipped with high-resolution cameras serve as valuable complements to traditional satellite remote sensing techniques. In this study, we propose a new spectral index called the visible band spectral slope index (VSSI) and integrate it with a triangular threshold segmentation algorithm to automatically extract features of <em>Cladophora</em> blooms from very-high-resolution UAV imagery. Comparative analysis with commonly used visible band indices such as the normalized green-red difference index (NGRDI), color index vegetation (CIVE), vegetation (VEG) index, visible-band difference vegetation index (VDVI), and red-green-blue floating algae index (RGB-FAI) demonstrated that the VSSI was most effective for detecting <em>Cladophora</em> blooms. The validation results revealed that the VSSI had the highest accuracy, achieving an F1 score of 0.89 and improving overall accuracy by 11.25 % to 36.92 % compared to other indices' accuracy levels. When compared with UAV images, both Sentinel-2 MSI and Landsat OLI images significantly overestimated the areas of <em>Cladophora</em> blooms by values of 110.65 % and 517.99 %, respectively. Moreover, UAV images captured at different times within the same day confirmed that wind speed and direction are crucial factors influencing dynamic changes in <em>Cladophora</em> blooms over short periods. This work provides a valuable reference for accurately mapping spatiotemporal dynamics of <em>Cladophora</em> blooms and effectively managing lake water environments.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103210"},"PeriodicalIF":5.8,"publicationDate":"2025-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144106102","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}

{"title":"A novel approach to mapping and monitoring land carbon sinks by combining remote sensing and biogeochemical modeling: A case study in Burkina Faso","authors":"Mohamed M.M. Ahmed ,&nbsp;Egor Prikaziuk ,&nbsp;Moritz Laub ,&nbsp;Annemarie L. Klaasse ,&nbsp;Lucas E. Ellerbroek","doi":"10.1016/j.ecoinf.2025.103174","DOIUrl":"10.1016/j.ecoinf.2025.103174","url":null,"abstract":"<div><div>Accurate and timely estimation of carbon sequestration in soil and forest biomass is crucial for applications such as carbon stock assessment, forest degradation monitoring, and climate change mitigation. Traditional methods such as field inventories, remote sensing, and biogeochemical models each have strengths and limitations, particularly in data-scarce regions. To address these challenges, this study integrates the light-use efficiency based ETLook model, which is driven by remotely sensed data, with the biogeochemical model DayCent, which is driven by management and weather data, to spatially model aboveground biomass and carbon sequestration. This novel approach aims to improve carbon sequestration estimates in a case study area in Burkina Faso, where ongoing political instability severely limits the availability of field data. In the absence of ground-truth data, we compare the outputs from DayCent and ETLook across time and space to build confidence in our estimates. Our findings indicate that, despite being driven by different input data, the DayCent model closely matches the aboveground biomass patterns observed in the ETLook model, with an r<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span> value of 0.81, a Kling-Gupta efficiency (KGE) of 0.77, low bias, and consistent seasonal patterns. Since ETLook lacks a soil carbon module, combining its Net Primary Productivity (NPP) and growth estimates with DayCent’s soil organic carbon (SOC) outputs provides a more robust estimate of total carbon sequestration than either model alone. Future work will focus on applying this hybrid approach across different ecological and geographical regions to evaluate its broader applicability.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"90 ","pages":"Article 103174"},"PeriodicalIF":5.8,"publicationDate":"2025-05-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144134881","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}

{"title":"Montrends: A Google Earth engine application for analysing species' habitat suitability over time","authors":"João Alírio ,&nbsp;Nuno Garcia ,&nbsp;João C. Campos ,&nbsp;Salvador Arenas-Castro ,&nbsp;Isabel Pôças ,&nbsp;Lia B. Duarte ,&nbsp;Ana Cláudia Teodoro ,&nbsp;Neftalí Sillero","doi":"10.1016/j.ecoinf.2025.103201","DOIUrl":"10.1016/j.ecoinf.2025.103201","url":null,"abstract":"<div><div>Human activities are impacting biodiversity worldwide. Biodiversity monitoring is essential to assess and support conservation status and trends. Remote sensing has played a crucial role in supporting biodiversity monitoring, but more intuitive and fast-processing tools are still required to improve biodiversity conservation. Herein, we present a Google Earth Engine (GEE) App called Montreds, which implements a biodiversity monitoring tool to measure trends in species habitat suitability over time by calculating ecological niche models (ENMs) with a time series of satellite products. The application is specific to Montesinho Natural Park/Nogueira Special Conservation Area, a protected area located in northeastern Portugal. The application calculates ENMs over time with MaxEnt for five taxa (vascular plants, amphibians, reptiles, birds, and mammals), using a time series of six Moderate-Resolution Imaging Spectroradiometer (MODIS) products between 2001 and 2023. Habitat suitability trends are estimated using the Mann-Kendall test. The Montrends' main output is a map for each modelled species with positive, negative, or null trends over time. If habitat suitability decreases monotonically over time, the trend is identified as negative. The application allows the users to select the species to be modelled, the temporal period, the number of model replicates, and the proportion of training and test records. The application runs the analyses intuitively in about a minute. Several results are displayed: the mean MaxEnt model over time and the Mann-Kendall trends for the whole study area, the species presences, the pixels with significant trends, and the species' occurrences in significant pixels. The application also provides the main MaxEnt outputs, including Area Under the Curve (AUC) values and variable contributions, plots of the global contributions of predictor variables over time, average trend values, and information on MaxEnt parameters. Decision-makers and conservation planners can use this application as a complementary tool for biodiversity monitoring and conservation.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103201"},"PeriodicalIF":5.8,"publicationDate":"2025-05-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144116283","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}

{"title":"A weather-driven mathematical model of Culex population abundance and the impact of vector control interventions","authors":"Suman Bhowmick ,&nbsp;Patrick Irwin ,&nbsp;Kristina Lopez ,&nbsp;Megan Lindsay Fritz ,&nbsp;Rebecca Lee Smith","doi":"10.1016/j.ecoinf.2025.103163","DOIUrl":"10.1016/j.ecoinf.2025.103163","url":null,"abstract":"<div><div>Even as the incidence of mosquito-borne diseases like West Nile Virus (WNV) in North America has risen over the past several decades, effectively modelling mosquito population density or abundance has proven to be a persistent challenge. It is critical to capture the fluctuations in mosquito abundance across seasons in order to forecast the varying risk of pathogen transmission from one year to the next. We develop a process-based mechanistic weather-driven Ordinary Differential Equation (ODE) model to study the population biology of both aquatic and terrestrial stages of mosquito population. The progression of mosquito lifecycle through these stages is influenced by different factors, including temperature, daylight hours, intra-species competition and the availability of aquatic habitats. In our work, weather-driven parameters are derived from a combination of laboratory research and data from the literature. In our model, we include precipitation data as a substitute for evaluating additional mortality in the mosquito population. We compute the <em>Basic offspring number</em> of the associated model and perform sensitivity analysis. Finally, we employ our model to assess the effectiveness of various adulticides strategies to predict the reduction in mosquito population. This enhancement in modelling of mosquito abundance can be instrumental in guiding interventions aimed at reducing mosquito populations and mitigating mosquito-borne diseases such as the WNV. This model could help optimise the timing of adulticide applications and evaluate the impact of multiple spray events within a short period.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103163"},"PeriodicalIF":5.8,"publicationDate":"2025-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144098353","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}

{"title":"Temporal generalization in evergreen leaf type classification using tailored Sentinel-2 composites","authors":"Peter Hofinger,&nbsp;Jan Dempewolf,&nbsp;Simon Ecke,&nbsp;Hans-Joachim Klemmt","doi":"10.1016/j.ecoinf.2025.103167","DOIUrl":"10.1016/j.ecoinf.2025.103167","url":null,"abstract":"<div><div>Large-scale forest ecosystem mapping relies critically on distinguishing deciduous and non-deciduous tree cover through advanced remote sensing technologies. Existing mapping approaches frequently suffer from spatial resolution limitations and temporal constraints. However, precise, high-fidelity forest cover characterizations are essential for forest management, ecological monitoring, and conservation planning. In this study we applied a novel methodology for classifying leaf types – evergreen versus deciduous – using Sentinel-2 multispectral satellite imagery at 10 meter resolution and machine learning, with the aim of strengthening the robustness of predictions and eliminating the need for retraining for unseen years when training on multi-year data. Key contributions include recursive feature elimination to identify the most relevant spectral bands and indices, and optimizing compositing methods to boost classification accuracy, balancing robustness and temporal detail. Eight machine learning models were tuned and trained on 16,162 tree crowns across 48 areas in Bavarian strict forest reserves (2019 to 2023) and validated with ForestGEO Traunstein Forest Dynamics Plot ground truth data (2018). We achieved an F1 score of 0.863 and an accuracy of 0.839 on the test area. Importantly, we found that model performance improved markedly with tree height, leading us to recommend our methodology for trees taller than 20 m. Results were benchmarked against the Copernicus High Resolution Layer Dominant Leaf Type product, with our top-performing model surpassing the Copernicus product in both metrics. This data-driven approach provides a scalable solution with temporal generalization, leveraging freely available satellite imagery and cloud-compute, aiding more effective forest management and environmental monitoring.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103167"},"PeriodicalIF":5.8,"publicationDate":"2025-05-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144089605","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}

{"title":"Association of IUCN-threatened Indian mangroves: A novel data-driven rule filtering approach for restoration strategy","authors":"Moumita Ghosh ,&nbsp;Sourav Mondal ,&nbsp;Rohmatul Fajriyah ,&nbsp;Kartick Chandra Mondal ,&nbsp;Anirban Roy","doi":"10.1016/j.ecoinf.2025.103164","DOIUrl":"10.1016/j.ecoinf.2025.103164","url":null,"abstract":"<div><div>Restoring biodiversity is crucial for ecological sustainability. This study introduces a novel data-driven rule-filtering framework that adopts domain knowledge of taxonomic distinctness and proposes a new metric, total taxonomic distinctness, to prioritize species selection for targeted restoration efforts. We extract and validate association rules to identify frequently co-occurring species and rank them based on total taxonomic distinctness. This structured approach ensures the selection of ecologically significant species that enhance biodiversity and ecosystem resilience. We apply this three-stage framework to Indian mangrove ecosystems, focusing on four IUCN Red List species: <em>Heritiera fomes</em>, <em>Sonneratia griffithii</em>, <em>Ceriops decandra</em>, and <em>Phoenix paludosa</em>. Our results indicate that taxonomically distinct species tend to co-occur more frequently, enhancing ecosystem resilience. Statistical validation using multiple hypothesis testing ensures the robustness of our findings. To assess the framework’s broader applicability, we extend our analysis to species presence-absence data from sacred groves in the laterite regions of eastern India. The results reinforce our previous findings, demonstrating frequent association patterns among taxonomically distinct species. This study provides actionable insights for ecological restoration, guiding species selection and co-planting strategies. The framework is adaptable across ecosystems, offering a scalable approach to biodiversity conservation.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103164"},"PeriodicalIF":5.8,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144069088","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}

{"title":"Regression analysis and artificial neural networks for predicting pine species volume in community forests","authors":"Wenceslao Santiago-García","doi":"10.1016/j.ecoinf.2025.103203","DOIUrl":"10.1016/j.ecoinf.2025.103203","url":null,"abstract":"<div><div>Volume prediction models are fundamental in forestry, as they support forest inventories, sustainable forest management strategies, and comprehensive environmental planning. The main objective of this study was to implement and compare two prominent approaches—regression and machine learning—for modeling whole-tree volume and stem volume in two <em>Pinus</em> species in community forests of southern Mexico. Destructive sampling provided data from 56 <em>P. patula</em> and 51 <em>P. pseudostrobus</em> trees, covering a wide range of diameters and heights. The regression approach employed seemingly unrelated nonlinear regression (NSUR) to fit simultaneous additive volume systems using both one- and two-variable models. In this approach, volume was modeled as a function of diameter at breast height (<em>d</em>) alone and as a function of both <em>d</em> and total tree height (<em>h</em>). Species and volume type were implicitly accounted for within the structure of the additive systems structure. For the machine learning approach, multilayer perceptron (MLP) artificial neural networks (ANNs) were trained using four input variables: diameter at breast height, total tree height, species, and volume type. These variables were explicitly incorporated into the ANN structure, enabling the model to learn complex, non-linear interactions. The ANN was optimized using L1 regularization and the Adam optimizer. The quantitative variables were diameter at breast height and total tree height, while the qualitative variables were species (<em>P. patula</em> and <em>P. pseudostrobus</em>) and volume type (whole-tree volume and stem volume), both coded as 1 and 0, respectively. The relative rank method was used to identify the most effective models based on goodness-of-fit statistics, including the coefficient of determination (R²), average absolute error (AAE), total relative error (TRE), average systematic error (ASE), and mean percent standard error (MPSE). The ANN approach consistently outperformed the regression model, achieving higher R² values and lower error metrics across all evaluations. Specifically, the ANN model reduced AAE, TRE, and ASE while maintaining biologically consistent predictions. This proposed ANN model represents a significant advancement in modeling both whole-tree and stem volume simultaneously and independently across different species, providing reliable and precise estimates. Given its ability to handle complex, non-linear relationships and its superior accuracy, we recommend the use of ANN as a practical tool in forestry applications, including forest resource evaluation and the development of sustainable forest management plans.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103203"},"PeriodicalIF":5.8,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144089606","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}

{"title":"Improved lightweight DeepLabV3+ for bare rock extraction from high-resolution UAV imagery","authors":"Pengde Lai ,&nbsp;Chao Lv ,&nbsp;Lv Zhou ,&nbsp;Shengxiong Yang ,&nbsp;Jiao Xu ,&nbsp;Qiulin Dong ,&nbsp;Meilin He","doi":"10.1016/j.ecoinf.2025.103204","DOIUrl":"10.1016/j.ecoinf.2025.103204","url":null,"abstract":"<div><div>Bare rock information extraction in karst regions is crucial for geological hazard monitoring and ecological assessment. However, in sparsely vegetated areas, bare rock exhibits similar spectral characteristics to surrounding land cover, and the boundaries are often indistinct, making it challenging for traditional classification methods to distinguish these transitional zones accurately. To address these challenges, this study proposes a bare rock extraction method based on an improved lightweight DeepLabV3+ model. MobileNetV2 is used as the backbone network, and the Channel Attention Module (CAM) and Spatial Attention Module (SAM) are introduced to enhance feature extraction capability. Results show the following: (1) When MobileNetV2 is used as the backbone of DeepLabV3+, the Accuracy, F1 score, and MIoU reach 97.39 %, 78.91 %, and 82.11 %, respectively, outperforming VGG16, Xception, SqueezeNet, and traditional segmentation models. (2) Applying the lightweight DeepLabV3+ model to bare rock identification in orthophoto imagery of the study area results in a bare rock rate error of approximately 5 %, demonstrating the practical applicability of the model. (3) After the introduction of the attention mechanism, the model's Recall, F1 score, and MIoU increased by 14.00 %, 8.37 %, and 5.62 %, respectively, remarkably enhancing identification completeness and boundary accuracy. Meanwhile, the improved model had a parameter count of 6.98 M and a computational complexity of 7.24G, achieving enhanced accuracy while maintaining computational efficiency. The research results can provide accurate bare rock information to support geological hazard monitoring and early warning, and offer new technical solutions for ecological restoration and risk assessment. (Data sets and code links: <span><span>https://figshare.com/articles/dataset/Bare_rock_dataset/28143443?file=53186633</span><svg><path></path></svg></span>).</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103204"},"PeriodicalIF":5.8,"publicationDate":"2025-05-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144089607","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}