Ecological Informatics最新文献

{"title":"Data-efficient self-supervised algorithms for fine-grained birdsong analysis","authors":"Houtan Ghaffari ,&nbsp;Lukas Rauch ,&nbsp;Paul Devos","doi":"10.1016/j.ecoinf.2026.103862","DOIUrl":"10.1016/j.ecoinf.2026.103862","url":null,"abstract":"<div><div>Research in bioacoustics, neuroscience, and linguistics often uses birdsong as a proxy to acquire knowledge across diverse areas. This requires audio models to annotate and parse the birdsong. Developing such models requires precise, syllable-level annotated training data. Therefore, automated methods that reduce annotation costs are in demand. This work presents a data-efficient birdsong annotator called Residual Multi-Layer Perceptron Recurrent Neural Network. It then presents a three-stage training pipeline for developing reliable birdsong syllable detectors with minimal annotation. The first stage is self-supervised learning from unlabeled data. Two of the most successful pretraining paradigms are explored, namely, masked prediction and online clustering. The second stage is supervised training with effective data augmentation to produce a robust frame-level syllable detector for each individual. The third stage is a semi-supervised post-training step that refines each individual’s model using unlabeled data. The effectiveness of this approach is demonstrated for the Canary song in extreme label-scarcity scenarios. From a signal-processing perspective, the Canary song exhibits one of the most challenging spectro-temporal patterns for algorithmic time-series annotation: rapid vocalizations, brief inter-syllabic intervals, fast and broadband frequency sweeps, and spectrally similar syllables that require fine-grained features to distinguish. Hence, a successful syllable detection algorithm for Canary also establishes a robust baseline for other birds. This methodological generalization is validated in a case study of Bengalese Finch song annotation. Finally, the potential of self-supervised embeddings is assessed for linear probing and unsupervised birdsong analysis.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103862"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185588","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":"Impacts of predicted climate and vegetation changes on pastoral systems: Insights from eleven Mediterranean case studies","authors":"Marco Bianchini ,&nbsp;Matteo Francioni ,&nbsp;Mohamed Tarhouni ,&nbsp;Abderrazak Tlili ,&nbsp;Jamila Msadek ,&nbsp;Maria Karatassiou ,&nbsp;Dimitrios Chouvardas ,&nbsp;Anna Sidiropoulou ,&nbsp;Afroditi Stergiou ,&nbsp;Jean-Paul Dubeuf ,&nbsp;Feliu López-i-Gelats ,&nbsp;Marta G. Rivera-Ferre ,&nbsp;Miquel Monés ,&nbsp;Antonio Franca ,&nbsp;Pasquale Arca ,&nbsp;Giovanni Altana ,&nbsp;Ante Ivanković ,&nbsp;Nazan Koluman ,&nbsp;Serap Göncü ,&nbsp;Lamis Chalak ,&nbsp;Paride D’Ottavio","doi":"10.1016/j.ecoinf.2026.103842","DOIUrl":"10.1016/j.ecoinf.2026.103842","url":null,"abstract":"<div><div>The Mediterranean basin is a climate change hotspot characterised by strong environmental heterogeneity and high vulnerability of pastoral systems. This study investigates how climate and vegetation dynamics may affect rangelands and pastoral systems across contrasting climatic zones. We analysed eleven case studies spanning temperate, semi-arid, and arid zones using an ensemble of three global circulation models under optimistic (SSP245) and pessimistic (SSP585) scenarios. Vegetation changes were modelled through spectral indices (NDVI and SAVI) using the MaxEnt algorithm to assess land suitability for forage resources. Results indicate a consistent warming trend (+3 to +7 °C by 2070), coupled with heterogeneous precipitation responses, with the strongest declines in semi-arid zones. Vegetation responses were highly context-dependent, ranging from shrubland expansion in temperate zones to increases in bare soil and discontinuous grasslands in semi-arid and arid systems. Some mountainous and northern areas may act as local refugia, showing milder changes or localised greening effects. These dynamics translate into differentiated impacts on pastoral systems. Water scarcity emerges as a key constraint, particularly in semi-arid and arid zones, while vegetation changes may affect forage availability, quality, and seasonality. Livestock systems could be increasingly exposed to heat stress, forage shortages, and changing disease dynamics, highlighting strong spatial heterogeneity in vulnerability. Therefore, adaptation strategies should be tailored to context conditions. Key strategies include improved water management, livestock mobility, opportunistic forage cultivation, and the use of locally adapted breeds. However, these may entail risks of maladaptation where they increase pressure on ecosystems or depend on socio-economic constraints. Overall, linking climate projections with system-level analyses is essential to support context-sensitive adaptation pathways and enhance the resilience of Mediterranean pastoral systems.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103842"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185584","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":"The LifeWatch ERIC Metadata Catalogue: A FAIR gateway to biodiversity and ecosystem research","authors":"Lucia Vaira ,&nbsp;Eftychia Tzafesta ,&nbsp;Majid Fiaz ,&nbsp;Houda Ben Salah ,&nbsp;Alberto Basset","doi":"10.1016/j.ecoinf.2026.103853","DOIUrl":"10.1016/j.ecoinf.2026.103853","url":null,"abstract":"<div><div>Biodiversity and ecosystem data follow a life cycle that includes several steps, such as planning, collection, quality improvement, and description and documentation through metadata. These are followed by preservation through curation and publication, as well as discovery and integration. The final stages involve analysis, which can lead to informed and effective decisions on conservation and restoration policies and strategies, as well as environmental management in general. This process involves a series of challenges, and the need to make data both fit-for-use and interoperable often represents one of the main limitations and barriers in (meta)data management. Despite ongoing efforts to promote common models by some communities, such as the European Network of enVironmental Research Infrastructures for environmental science, the harmonisation of vocabularies and metadata across Research Infrastructures remains a persistent challenge, locking data away. For this purpose, LifeWatch ERIC, the e-Science European infrastructure for biodiversity and ecosystem research, developed the LifeWatch ERIC Metadata Catalogue, a comprehensive, standards-based information management system that ensures records are Findable, Accessible, Interoperable, and Reusable (FAIR). In doing so, it enables structured and standardised information to be openly available, promoting reuse, sharing, and integration. This paper presents the LifeWatch ERIC Metadata Catalogue by describing its conceptual architecture and its core functionalities.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103853"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148167990","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":"An integrated framework for monitoring forest recovery process after disturbance based on Landsat time series images and deep learning methods","authors":"Ke Xu ,&nbsp;Jinzhuo Wu ,&nbsp;Wenshu Lin ,&nbsp;Ning Zhang ,&nbsp;Yifan Xu ,&nbsp;Yang Long","doi":"10.1016/j.ecoinf.2026.103867","DOIUrl":"10.1016/j.ecoinf.2026.103867","url":null,"abstract":"<div><div>The increasing frequency and intensity of forest disturbances, coupled with varying recovery capacities of forest ecosystems, underscore the importance of understanding the full spectrum of disturbance and subsequent recovery dynamics. Previous research has primarily focused on binary forest disturbance mapping, with limited attention given to dynamic disturbance-recovery processes. In this study, we generated a synthetic disturbance-recovery dataset by analyzing the trajectory characteristics of LandTrendr-fitted Normalized Burn Ratio (NBR) time series and capturing multitemporal patterns before, during, and after disturbances. This dataset was used to train an InceptionTime classification model, which was subsequently applied to the LandTrendr-fitted time series to classify different disturbance-recovery processes. Finally, three complementary resilience indicators including Disturbance Magnitude (DM), Year-on-Year Average (YrYr), and State Transition Rate (STR) were calculated to assess forest ecosystem resilience. Results showed that the InceptionTime model achieved higher classification accuracy on the synthetic dataset (OA = 96%), and the overall accuracy of the resulting disturbance-recovery classification map was 91% (±3%). The classification maps indicated that 44.8% disturbance area in the study area exhibited incomplete recovery (DR2), 42.9% completely recovered (DR1), while only a small proportion (6.8%) showed no recovery (DN). The DM value gradually declined over time, while the YrYr value exhibited a subtle upward trend, indicating a progressive enhancement in forest ecosystem resilience. Our study provides a framework for characterizing forest disturbance-recovery processes and assessing ecosystem resilience in the absence of in-situ data, thereby offering valuable insights for forest resource management.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103867"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185181","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":"GIS-based modeling of the complex bioclimatic belts for investigating natural vegetation distribution knowledge within protected areas","authors":"Cuong Dang-Hung ,&nbsp;Hoi Dang Nguyen ,&nbsp;Tam Minh Pham","doi":"10.1016/j.ecoinf.2026.103834","DOIUrl":"10.1016/j.ecoinf.2026.103834","url":null,"abstract":"<div><div>Bioclimatic classification provides a framework for understanding how climatic factors shape the distribution of natural vegetation through biologically relevant categories. Especially in tropical regions, the complexity of bioclimatic belts reflects major gradients in ombrotypes, and thermotypes, leading to diverse terrestrial ecosystems. However, large-scale GIS-based modeling of bioclimate units remains challenging due to limited observational data and spatial uncertainties. To address these issues, we developed a high-resolution (1:100,000) bioclimatic map of Gia Lai province, Vietnam, using a set of parameters and indices derived from the Worldwide Bioclimatic Classification System. Temperature and precipitation data from the WorldClim dataset (1 km resolution) were downscaled using the Inverse Distance Weighting (IDW) interpolation method and validated with records from three meteorological stations (1980–2020). Our result identified nine isobioclimatic units across the study area. The diagnosis of bioclimatic belts reveals drier conditions in the southwest and more humid conditions in the northeast, corresponding to distinct vegetation patterns. These findings demonstrate the strong influence of bioclimatic variability on potential natural vegetation distribution and provide a robust basis for predicting vegetation composition and ecosystem responses to climate. This approach supports data-driven strategies for forest restoration and biodiversity conservation in tropical regions.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103834"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185637","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 machine learning framework for radon forecasting with uncertainty quantification in the Campi Flegrei volcanic area","authors":"Michele Di Giovanni ,&nbsp;Fabrizio Ambrosino ,&nbsp;Flora Giudicepietro ,&nbsp;Walter De Cesare ,&nbsp;Mariagabriella Pugliese ,&nbsp;Giuseppe La Verde ,&nbsp;Francesco A.N. Palmieri ,&nbsp;Giovanni Di Gennaro ,&nbsp;Carlo Sabbarese","doi":"10.1016/j.ecoinf.2026.103864","DOIUrl":"10.1016/j.ecoinf.2026.103864","url":null,"abstract":"<div><div>This study presents a machine learning framework for short-term forecasting of radon activity concentration in the Campi Flegrei volcanic area. We integrate fourteen years of radon, meteorological, and seismic observations through a dedicated preprocessing pipeline that addresses missing values, uneven sampling, and aggregation into 3-hour intervals. Two model families are compared: a tree-based predictor (XGBoost) and a sequence-based predictor (LSTM). Conformal prediction is employed as an uncertainty layer to complement deterministic forecasts with predictive intervals. Point-prediction performance is quantified in terms of Root Mean Square Error (RMSE) and Pearson correlation: the best one-step predictor achieves a normalised RMSE of about 15% (normalised by the mean concentration) together with a high correlation coefficient (<span><math><mrow><mi>ρ</mi><mo>≈</mo><mn>0</mn><mo>.</mo><mn>95</mn></mrow></math></span>). Uncertainty quantification relies on Conformal Prediction (CP), and is summarised by the empirical Prediction-Interval Coverage Percentage (PICP) with respect to the nominal level, together with interval sharpness measures; for a representative configuration, PICP remains close to the nominal level (around 90%). Finally, recursive free-run experiments highlight the expected horizon-dependent degradation: errors grow with the forecast horizon for both predictors, with XGBoost attaining slightly lower mean error while LSTM shows greater robustness through reduced variability across rollouts at medium and long horizons. Overall, the proposed framework integrates well-established methodologies for radon forecasting and uncertainty-aware evaluation, enhancing the reliability of environmental time-series forecasting and providing a reproducible tool.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103864"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148186225","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":"Nonlinear driving mechanisms of social-ecological factors on land carbon sources and sinks in urban agglomeration considering spatial effects: An explainable geospatial machine learning model","authors":"Xiaojie Guo ,&nbsp;Lei Yuan ,&nbsp;Yong Wang ,&nbsp;Shengjun Zhu","doi":"10.1016/j.ecoinf.2026.103843","DOIUrl":"10.1016/j.ecoinf.2026.103843","url":null,"abstract":"<div><div>Revealing the impact of socio-ecological factors on land carbon sources/sinks is the key to balancing ecological protection and economic development under the carbon neutrality goal. However, existing studies are limited by the separation of source-sink perspectives, insufficient dynamic interaction analysis, and the difficulty in simultaneously depicting spatial effects and nonlinear relationships, which restricts the in-depth understanding of the driving mechanism. Based on estimated land carbon emissions (LCE) and land carbon sinks (LCS) in Yunnan central urban agglomeration (YNUA), this study incorporated geographic coordinates (GEO) as spatial features and applied an explainable geospatial machine learning (XGeoML) model that integrates LightGBM and GeoShapley. This model was used to systematically examine the nonlinear driving mechanisms from four perspectives: importance and directionality, threshold effects, interaction effects, and spatial effects. The results revealed that (1) LCE continued to rise but at a decelerating rate, while LCS showed an overall declining trend with fluctuation. Both exhibited significant spatial clustering. (2) Human activities, climate, and vegetation factors jointly explained over 65% of the spatiotemporal variation in LCE and LCS, with the independent influence of GEO surpassing that of most driving factors. (3) The influence of socioeconomic factors varied with the stage of urban development. Natural ecological factors exhibited multi-threshold nonlinear responses, and landscape pattern optimization supported emission reduction and sink enhancement. Significant dynamic interactions were observed among the factors. (4) From the local scale, GEO not only directly shaped the spatial pattern of LCE and LCS but also modulated the intensity and direction of other factors, demonstrating a clear spatial regulatory effect. This study emphasizes the necessity of incorporating nonlinearity and spatial effects into carbon management and provides a scientific basis for precise regional emission reduction and sink enhancement.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103843"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185733","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 height-above-river metrics and machine learning to model riparian vegetation distribution and reach-scale restoration potential","authors":"Clancy R. McConnell ,&nbsp;James H. Thorne ,&nbsp;Steven E. Greco","doi":"10.1016/j.ecoinf.2026.103869","DOIUrl":"10.1016/j.ecoinf.2026.103869","url":null,"abstract":"<div><div>Rivers worldwide have been geomorphologically degraded by damming, channelization, and floodplain disconnection, reducing the extent of riparian habitat and complicating restoration planning. We developed and evaluated a novel, system-wide workflow that integrates height-above-river (HAR) metrics with a random forest classifier to model riparian vegetation distribution, assess geomorphic condition, and estimate restoration potential. This approach is, to our knowledge, the first to generate a probabilistic land-cover model based solely on HAR and to use this relationship to support reach-scale restoration prioritization. Applied to a regulated river in California, the model predicted six land-cover classes from HAR with 88% overall accuracy (κ = 0.76), capturing clear vertical patterns in vegetation distribution. These relationships were used to define six elevation-based zones (aquatic, core riparian, marginal riparian, transition, valley oak, and out-of-channel) that represent functional differences in hydrologic connectivity and vegetation structure. Using these zones, 25 geomorphologically uniform reaches were delineated and classified by degradation based on the relative extent of low-elevation riparian area versus elevated, disconnected floodplain surfaces. Restoration potential, defined as the area of core riparian habitat that could be created through floodplain lowering and channel narrowing, varied from 2.3 to 33.9 ha per reach. Across the study area, restoration could increase core riparian habitat by 186% (197 ha). Because the workflow relies only on LiDAR-derived topography and aerial imagery, which are becoming widely available, it is intended as a system-scale planning tool for use by land managers prior to field-based geomorphic design assessments, such as those used in restoration implementation projects.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103869"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148241940","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":"Application and field validation of machine and deep learning models for predicting nitrogen losses in rice paddies with fertilizer deep-placement","authors":"Yun-Gu Kang ,&nbsp;Jun-Yeong Lee ,&nbsp;Jiwon Choi ,&nbsp;Jun-Ho Kim ,&nbsp;Taek-Keun Oh","doi":"10.1016/j.ecoinf.2026.103845","DOIUrl":"10.1016/j.ecoinf.2026.103845","url":null,"abstract":"<div><div>Mitigating nitrogen (N) loss from rice paddies is essential for enhancing rice productivity while reducing environmental impacts. This study evaluated two machine learning (ML) models (i.e., random forest and extreme gradient boosting; XGBoost) and one deep learning (DL) model (i.e., long short-term memory; LSTM) for predicting ammonia (NH₃) and nitrous oxide (N₂O) emissions under varying fertilizer deep-placement (FDP) conditions. Models were developed using data from a controlled soil incubation experiment (<em>n</em> = 285) designed to simulate flooded paddy conditions. The models were subsequently validated using an independent field data (<em>n</em> = 42) collected during a 100-day rice cultivation period. Input features included fertilization depth, days after fertilization, soil properties, inorganic N content, and atmospheric temperature. Model performance was assessed using four metrics, including coefficient of determination (R²) and root mean squared error (RMSE). Under controlled conditions, the ensemble ML models showed stable and reliable predictive performance, with low variability across cross-validation. In contrast, under field conditions characterized by greater environmental heterogeneity and temporal variability, the LSTM model demonstrated superior predictive performance, explaining up to 81% and 97% of the variability in N₂O and NH₃ emissions, respectively, under field conditions. The LSTM model also provided reasonable estimates of cumulative NH₃ losses, achieving 94–97% agreement with observations. These results suggest that incorporating temporal dependencies is beneficial for capturing N loss dynamics under open-field conditions. Overall, the findings indicate that LSTM-based approaches can serve as a complementary tool for evaluating FDP effectiveness and supporting improved fertilization management in rice paddies.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103845"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185598","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 spatiotemporal method for NDVI reconstruction based on differential smoothness","authors":"Chen Chen ,&nbsp;Wang Fu ,&nbsp;Tiejian Li ,&nbsp;Xindong Yao ,&nbsp;Xingyi Ou ,&nbsp;Bellie Sivakumar ,&nbsp;Jiaye Li ,&nbsp;Jing-Cheng Han ,&nbsp;Yuantao Gu ,&nbsp;Jiahua Wei","doi":"10.1016/j.ecoinf.2026.103856","DOIUrl":"10.1016/j.ecoinf.2026.103856","url":null,"abstract":"<div><div>The Normalized Difference Vegetation Index (NDVI) is one of the most widely used remote sensing indicators for assessing vegetation status, with applications across a broad range of studies including phenology, ecology, and hydrology. However, NDVI data are usually degraded by clouds, snow cover, and other factors. In this study, we propose a novel spatiotemporal method—the Time-varying Graph Signal Reconstruction (TGSR) method—for the reconstruction of low-quality NDVI data. This method regards spatiotemporal NDVI data as time-varying graph signals and reconstructs them by maximizing the differential smoothness of the signal. We applied the TGSR method to sample patches (128<span><math><mo>×</mo></math></span>128 pixels) from MODIS/Terra Vegetation Indices 16-Day L3 Global 500 m Grid products across East Asia, and quantitatively compared the results with those from two benchmark methods: the Savitzky-Golay (SG) filter method and the Spatiotemporal Tensor (ST-Tensor) method. Results demonstrate that the TGSR method outperforms both benchmark methods in capturing temporal dynamics and preserving spatial structure of NDVI data, achieving an average root mean square error of 0.0302, compared to 0.0354 for SG and 0.0331 for ST-Tensor. The TGSR method exhibits no bias toward noise type—whether positively biased, negatively biased, or missing data—and shows strong robustness against varying noise intensities, performing well even with only 15% reliable data. Moreover, the TGSR method effectively addresses challenges related to spatiotemporal continuous gaps and land cover changes. The present findings highlight the strong potential of the TGSR method for generating high-quality NDVI datasets and its applicability to similar remote sensing time-series products.</div></div>","PeriodicalId":51024,"journal":{"name":"Ecological Informatics","volume":"96 ","pages":"Article 103856"},"PeriodicalIF":7.3,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148185180","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}