Ecological Informatics最新文献

{"title":"Mapping individual tree crowns to extract morphological attributes in urban areas using unmanned aerial vehicle-based LiDAR and RGB data","authors":"Geonung Park ,&nbsp;Bonggeun Song ,&nbsp;Kyunghun Park","doi":"10.1016/j.ecoinf.2025.103165","DOIUrl":"10.1016/j.ecoinf.2025.103165","url":null,"abstract":"<div><div>Mapping individual tree crowns (ITCs) along with their morphological attributes provides foundational variables for estimating functions, such as thermal stress and carbon emissions, within the urban environment. However, to calculate morphological attributes, it is necessary to delineate ITCs, and applying the watershed segmentation (WS) algorithm, commonly used in forests, to urban environments presents challenges due to the reliance on single-band data and complexity of heterogeneous urban elements. Additionally, deep learning (DL) models, which excel in image analysis, are constrained by the labor-intensive label generation process. This study introduces a novel framework integrating machine learning (ML)- and DL-based approaches to map ITCs and extract the morphological attributes in urban areas using unmanned aerial vehicles (UAVs). Using ML-based approaches, we conducted object-based image analysis to optimize the use of the WS algorithm because applying WS directly to urban environments, as in forests, overestimated the ITC size by 151.37 %. This approach also improved the effectiveness of the DL model, Mask R-CNN, by resolving challenges in label generation. Mask R-CNN delineated ITCs with an accuracy of 0.942, suggesting its robustness in handling the heterogeneity of tree arrangements. The results demonstrate that the proposed framework is applicable for use in urban areas globally that have similar ecological conditions. ITCs with morphological attributes provide fundamental variables for evaluating ecological functions, which are scalable for improving urban environmental planning. However, UAV data may face time and cost limitations as the monitoring coverage expands, which should be considered when applying this framework.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103165"},"PeriodicalIF":5.8,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143863838","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":"Estimation of mangrove heights and aboveground biomass using UAV-LiDAR, Sentinel-1 and ZY-3 stereo images","authors":"Bolin Fu ,&nbsp;Yingying Wei ,&nbsp;Linhang Jiang ,&nbsp;Hang Yao ,&nbsp;Xiaomin Li ,&nbsp;Yanli Yang ,&nbsp;Mingming Jia ,&nbsp;Weiwei Sun","doi":"10.1016/j.ecoinf.2025.103160","DOIUrl":"10.1016/j.ecoinf.2025.103160","url":null,"abstract":"<div><div>Mangroves are crucial blue carbon ecosystems that are essential for promoting sustainable global development. Tree height is a key indicator of mangrove health; however, accurately estimating mangrove height in complex coastal environments is challenging. In this study, we constructed mangrove height inversion models using multiple types of remote sensing data and machine learning algorithms (partial least squares regression (PLSR), random forest (RF), and mixture density network (MDN)). We evaluated the performance of UAV-LiDAR point clouds, ZY-3 stereo images, and Sentinel-1 polarimetric and interferometric data in mangrove height inversion, and explored the accuracy differences among the dominant species. We also estimated the aboveground biomass of different dominant mangrove species to better understand their ecological functions and health conditions. The results showed the following: (1) The canopy height model and height variables of the LiDAR point clouds, DVI and near-infrared bands of the ZY-3 stereo images, and polarimetric decomposition parameters of the Sentinel-1 SAR images were more sensitive to mangrove heights. (2) The LiDAR point clouds and Sentinel-1 SAR images achieved the highest inversion accuracy when using the RF algorithm, with R² values of 0.875 and 0.685, respectively. The ZY-3 stereo images based on MDN obtained the optimal inversion results (R² = 0.719), with an improvement ranging from 0.143 to 0.198 when compared to the PLSR and RF algorithms. (3) <em>Avicennia marina</em> was associated with the highest estimation accuracy (R² = 0.897) compared to the other dominant mangrove species. <em>Aegiceras corniculatum</em> and <em>Avicennia marina</em> were associated with the highest inversion accuracy within the height range of 2–3 m (R² = 0.925, R² = 0.814, respectively), whereas <em>Kandelia candel</em> yielded the optimal inversion results at the height range of 1–2 m (R² = 0.652). (4) The aboveground biomass of <em>Aegiceras cornicatum</em> and <em>Kandelia candel</em> ranged from 20.176 to 103.164 Mg/ha and 132.019 to 719.226 Mg/ha, respectively, and the aboveground biomass of <em>Avicennia marina</em> was mainly distributed within the range of 169.916 to 803.204 Mg/ha. Our study provides a reference for monitoring the heights and health of mangroves, as well as their protection and development.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103160"},"PeriodicalIF":5.8,"publicationDate":"2025-04-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143863842","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":"Early summer warming amplification threats towards sustainable development goals (SDGs) in India","authors":"Susanta Mahato ,&nbsp;Swades Pal ,&nbsp;P.K. Joshi ,&nbsp;Andreas Matzarakis ,&nbsp;Paolo Tarolli ,&nbsp;Vicky Anand","doi":"10.1016/j.ecoinf.2025.103156","DOIUrl":"10.1016/j.ecoinf.2025.103156","url":null,"abstract":"<div><div>This study investigates the anomalous surge in early summer Land Surface Temperature (LST) in India and its potential repercussions on various sectors, such as food security, energy resources, and public health. The research also assesses the implications of the accomplishment of the Sustainable Development Goals (SDGs) throughout the early summer. Analyzing data from 2001 to 2022, the findings reveal that early summer LST was notably increased, with daytime temperatures exceeding mean LST by 3.5–4.14 °C and nighttime temperatures by 0.83 to 2.41 °C. Anomalous positive Standard Anomaly (StA) deviations were prevalent in north-west, central northeast, west-central, and hilly regions during the day. Trend analysis indicated varying StA responses across six homogeneous monsoon regions, with an overall positive trend observed in most areas. Surprisingly, Sea Surface Temperature (SST), which typically influences summer heating, was not the primary driver in 2022. Instead, a prolonged rain deficit in significant parts of India was identified as the cause. Regression analysis between StA and crop yields showed statistically insignificant associations for most production regions, except for a detrimental impact on winter crop yields. Energy deficits of up to 15 % were recorded in heat-affected states. The study also considered potential health issues arising from summer warming. These cumulative effects pose significant challenges to India's economic growth. The study assesses mitigation strategies discussed at the COP27 summit to address early summer warming. The findings provide valuable insights for developing preparedness and resilience plans to mitigate these issues.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103156"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855329","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":"Estimation of phytoplankton community composition from satellite data using a fuzzy and probabilistic combination model in mountainous reservoirs: A case of Huating Lake in spring and summer","authors":"Yixuan Qiu ,&nbsp;Zhongya Fan ,&nbsp;Huiyun Feng ,&nbsp;Yutao Wang ,&nbsp;Dan Li ,&nbsp;Wencai Wang ,&nbsp;Ruting Huang ,&nbsp;Jingang Jiang","doi":"10.1016/j.ecoinf.2025.103153","DOIUrl":"10.1016/j.ecoinf.2025.103153","url":null,"abstract":"<div><div>Although remote sensing has become a common tool for monitoring mountainous reservoirs, studies on the detection of phytoplankton community compositions (PCCs) remain insufficient. Based on satellite and field data, we developed a mathematical model incorporating fuzzy logic and probabilistic methods to directly estimate the biomass of seven different phytoplankton species in Huating Lake. Water surface temperature (WST) and chlorophyll-a concentration ([Chl-a]) were selected as input parameters for this model. The WST data were processed using a single-channel algorithm that combined the brightness temperature conversion model and land surface emissivity algorithm. Inversion of [Chl-a] was conducted using an empirical approach to compare the four models developed for the two sensitive reflectance bands. The [Chl-a] values obtained from these models were significantly correlated with the field data (<em>R</em> &gt; 0.8). The optimal model was selected based on validation results. After obtaining the inversion results for the WST and [Chl-a], we applied a fuzzy probabilistic model to determine the PCCs in Huating Lake from 2013 to 2023. A comparison with the measured data confirmed that this method reliably estimated PCC biomass (<em>R</em> &gt; 0.65). However, the modeling accuracy was not particularly high for Bacillariophyta and Euglenophyta with high biomass. We analyzed the spatial and temporal distribution of PCCs in Huating Lake over 10 years from 2013 to 2023 and found that the results were reasonable. The results demonstrate that the fuzzy probabilistic approach offers a novel methodology for estimating the biomass of seven phytoplankton species. This method facilitates the expansion of remote-sensing technology for monitoring PCC changes in mountainous reservoirs and provides scientific data support for understanding algal bloom mechanisms and developing prevention strategies.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103153"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143887221","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":"Capturing plant functional traits in coastal dunes using close-range remote sensing","authors":"Giacomo Trotta ,&nbsp;Marco Vuerich ,&nbsp;Elisa Petrussa ,&nbsp;Edoardo Asquini ,&nbsp;Paolo Cingano ,&nbsp;Francesco Boscutti","doi":"10.1016/j.ecoinf.2025.103159","DOIUrl":"10.1016/j.ecoinf.2025.103159","url":null,"abstract":"<div><div>Coastal dunes are dynamic ecosystems characterized by steep environmental gradients that impose significant stress on plant communities. These stressors, such as salinity, drought, and nutrient-poor soils, create a mosaic of plant communities with strong functional trait identity. Several studies have focused on plant functional responses to environmental conditions, but a gap remains in connecting plant functional traits to large-scale ecological processes through remote sensing. We studied a dune plant community (a total of 17 species) and the ecosystem key species <em>Cakile maritima</em> Scop. to explore how remote sensing-derived vegetation indices correlate with plant growth and specific physiological and morphological leaf traits, including specific leaf area, leaf dry matter content, and flavonoid concentration. We introduced a close-range approach using multispectral imaging to capture high-resolution (1.3 mm/px) data on plant functional traits in coastal dune ecosystems overcoming the limitations of broader-scale remote sensing methods which often suffer from lower spatial resolution and interference from non-vegetated areas. By semi-automatically identifying regions of interest for each species and eliminating background noise, we acquired accurate multispectral signatures that represent plant responses and highlight ecological processes of the key species and the broader community. We observed traits to be stronger than plant growth in explaining the variance of multispectral indices, with leaf flavonoids showing the highest contribution to plant spectral signature.</div><div>We demonstrated the effectiveness of close-range multispectral imaging in linking plant traits to ecological processes, with significant implications for upscaling plant responses to environmental variable across larger spatial scales. Furthermore, the research outlines practical guidelines for collecting and processing close-range multispectral data, offering a valuable new tool for and accurate field monitoring of ecosystem processes and plant functions.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103159"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860501","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":"Time series modelling spatiotemporal changes in Biogeoclimatic ecosystem classification (BEC) zones between 1997 and 2019 in West-Central British Columbia, Canada","authors":"Ilythia D. Morley ,&nbsp;Kevin Hanna ,&nbsp;Chris T. Darimont ,&nbsp;Mathieu L. Bourbonnais ,&nbsp;Ilythia D. Morley","doi":"10.1016/j.ecoinf.2025.103155","DOIUrl":"10.1016/j.ecoinf.2025.103155","url":null,"abstract":"<div><div>Understanding the spatial extent and temporal variability of ecosystem processes is essential for contextualizing land use and land cover change due to disturbance. In this study, we apply an advanced time series modelling method to assess and map ecosystem change and characterize ecosystem cover in west-central British Columbia, Canada. We couple Biogeoclimatic Ecosystem Classification (BEC) zone data with metrics derived from Landsat imagery to model how biogeoclimatic ecosystem cover, interpreted as an indicator of shifting vegetation seasonality, varies over a broad spatiotemporal scale. To do so, we apply the Time-Weighted Dynamic Time Warping (TWDTW) time series modelling approach by relating the spectral characteristics of Landsat data and derived indices from 1997 to 2019. Results highlight important transitions between biogeoclimatic ecosystem classes, with a transition of the interior Douglas-fir Dry to the montane-spruce Dry and the Sub-Boreal Pine to the Spruce zone Dry zones in response to large wildfires in 2003 and 2009. The assessment of ecosystem change across broad spatial and temporal scales is important for assessing the cumulative impacts of changes across highly variable landscapes on essential landscape services.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103155"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143860502","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":"Using fishery-related data, scientific expertise, and machine learning to improve marine habitat mapping in northeastern Mediterranean waters","authors":"Loukas Katikas ,&nbsp;Sofia Reizopoulou ,&nbsp;Paraskevi Drakopoulou ,&nbsp;Vassiliki Vassilopoulou","doi":"10.1016/j.ecoinf.2025.103154","DOIUrl":"10.1016/j.ecoinf.2025.103154","url":null,"abstract":"<div><div>Marine habitat mapping is an essential tool for planning conservation efforts and sustainable management of marine activities. High spatial resolution in marine habitat maps is of utmost importance, as it may encompass more detail in imagery and reveal important biotopes. This level of detail supports directing monitoring and analysis efforts for effective implementation of European Union (EU) environmental policies and provides more relevant advice for robust decision-making under sectorial policies (e.g., the Common Fisheries Policy) and more integrated policies (e.g., marine spatial planning). In this study, sea bottom type data recorded during national monitoring of commercial fishing vessel operations and fishery surveys in the Greek Seas were used. These data were then assigned to the EU EMODnet seabed habitats using local ecological knowledge. Two machine-learning algorithms, i.e., random forest classifier (RFC) and gradient boosting classifier, were trained on the entire national-scale dataset and subsequently applied to assess their performance in predicting habitat types in the Saronikos Gulf (regional scale) using various environmental factors as predictors. The borderline synthetic minority oversampling technique was applied to manage inherent data class imbalances. A validation dataset and georeferenced data from previous studies were used to compare the accuracy and predictive performance of the models. Using this approach, the Saronikos Gulf was enriched with five more habitat types than visualised in the EMODnet portal, which also filled habitat gaps in areas where no data existed. Results from application of the RFC-Borderline Smote (BS) model (62 % accuracy, 0.51 kappa score) were then used to address conservation planning commitments recently made by the Greek government. The vast majority of marine seabed priority habitats in the study area appeared to fall outside the borders of the current Natura 2000 sites, which served as the baseline for the declared trawl bans in Greek waters, following the provisions of the EU Marine Action Plan.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103154"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855330","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":"Automated near real-time monitoring in ecology: Status quo and ways forward","authors":"Anna Marie Davison ,&nbsp;Koen de Koning ,&nbsp;Franziska Taubert ,&nbsp;Jan-Kees Schakel","doi":"10.1016/j.ecoinf.2025.103157","DOIUrl":"10.1016/j.ecoinf.2025.103157","url":null,"abstract":"<div><div>In the current epoch of rapid biodiversity decline, monitoring of ecosystems and the species which inhabit them has become increasingly important. A near real-time approach to ecological monitoring facilitates decision making and timely interventions within rapidly changing systems. Despite fast-paced technological advancements making the automated workflows required for near real-time ecological monitoring possible, their use is highly limited and there is yet to be a review of the current capacity for their creation. This paper summarises the current methods and technologies which could be used to create such workflows and the considerations for establishing them in decision-making systems. We identify key barriers to the adoption of a NRT approach across geographies and different fields of study in ecology. We also highlight the need to work collaboratively with technologists and stakeholders to establish efficient and long-lasting NRT workflows which can inform evidence-based decision making.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103157"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143873952","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":"Exploring new mangrove horizons: A scalable remote sensing approach with Planet-NICFI and Sentinel-2 images","authors":"Adam Irwansyah Fauzi ,&nbsp;Markus Immitzer ,&nbsp;Clement Atzberger","doi":"10.1016/j.ecoinf.2025.103152","DOIUrl":"10.1016/j.ecoinf.2025.103152","url":null,"abstract":"<div><div>Mangroves offer massive ecosystem services ranging from coastal protection, and wildlife habitat to carbon sequestration. This makes them an integral part of tropical developing countries' strategies to pursue climate neutrality targets. In this respect, the advanced development of big data, machine learning, and cloud computing in remote sensing provides a huge opportunity to explore this ecosystem and to provide scalable monitoring solutions. This study aims to discover new potential mangrove areas, focusing on far-off and under-monitored locations along the coasts and rivers of Indonesia, using a precise, practical, and scalable remote sensing approach via Google Earth Engine. To demonstrate the potential of our approach, we selected Lampung province, Indonesia as the study area, which has varied taxonomic, topographical, bathymetrical, and oceanographical characteristics. The methodology includes defining mapping zones using coastline, river, and elevation data. The satellite image processing is based on integrating Planet-NICFI and Sentinel-2 images using the Simple Non-Iterative Clustering (SNIC) segmentation and Random Forest (RF) classifier. Our classification with F1 score of 0.95 successfully mapped 10,290 ha of mangroves, with coastal mangroves contributing 6058 ha and riverine mangroves another 4250 ha. Importantly, this study discovered 1714 ha of previously unknown mangroves, equivalent to 18.55 % of the official area. These new areas are dominated by nypa palm, a native species that contributes to bioeconomy. This study contributes to refining carbon sequestration baselines and highlights the scalability of a national-level implementation to support progress towards net-zero emissions goals. The method can readily be deployed to other mangrove areas.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"88 ","pages":"Article 103152"},"PeriodicalIF":5.8,"publicationDate":"2025-04-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143855331","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":"Robust real-time detection of right whale upcalls using neural networks on the edge","authors":"Matthew D. Hyer ,&nbsp;Austin T. Anderson ,&nbsp;David A. Mann ,&nbsp;T. Aran Mooney ,&nbsp;Nadège Aoki ,&nbsp;Frants H. Jensen","doi":"10.1016/j.ecoinf.2025.103130","DOIUrl":"10.1016/j.ecoinf.2025.103130","url":null,"abstract":"<div><div>Animals worldwide are facing ecological pressures from global climate change and increasing anthropogenic activities. To transition to a renewable energy future, extensive offshore wind development is planned globally. In the North Atlantic, future development sites overlap with the migratory range of critically endangered North Atlantic right whales (NARW) and will lead to increased risk of ship strikes, pile driving impacts, and other population risks. New methods to accurately detect cetaceans and provide real-time feedback for mitigation will be increasingly important to enact sustainable management actions to facilitate the recovery of the NARW. Recent developments in acoustic event detection made possible by deep learning have shown significantly improved detection performance across many different taxa, but such models tend to be too computationally expensive to run on existing wildlife monitoring platforms. Here, we use model compression techniques combined with an autonomous acoustic recording platform integrating an ESP32 microcontroller to bring real-time detection with deep learning to the edge. We test if edge-based inference using a compressed network running on a microprocessor entails significant performance loss and find that this loss is negligible. We leverage large, open-source datasets of noise from the NOAA SanctSound project for generating semi-synthetic training datasets that encourage model generalization to novel noise conditions. Our compressed model achieves improved performance across all tested recording sites in the Western North Atlantic Ocean, demonstrating that deep learning powered wildlife monitoring solutions can provide reliable real-time data for mitigation of human impacts and help ensure a sustainable green energy transition.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"89 ","pages":"Article 103130"},"PeriodicalIF":5.8,"publicationDate":"2025-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143873953","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}