Pub Date : 2026-01-24DOI: 10.1016/j.envsoft.2026.106894
Rezgar Arabzadeh, Jonathan Romero-Cuellar, James Craig, Bryan Tolson, Robert Chlumsky
Bayesian calibration of hydrologic models effectively addresses parameter uncertainties and improves predictions, but the joint inference of hydrologic and error model parameters often suffers from slow convergence due to high-dimensional interactions. To overcome this, a surrogate-aided error model is introduced that decouples their inference. This method uses support vector regression (SVR) as a surrogate, to estimate error model parameters conditioned on hydrologic parameters. This approach accelerates convergence (requiring 50 % fewer samples) and improves predictive accuracy, consistently improving or maintaining Continuous Ranked Probability Scores across a range of test models. These advantages are demonstrated through an application to the GR4J model across 12 MOPEX watersheds. The reduced computational demand makes this particularly valuable for large-scale hydrologic modeling when computational resources are limited.
{"title":"A surrogate-aided approach for accelerated Bayesian calibration of hydrologic models","authors":"Rezgar Arabzadeh, Jonathan Romero-Cuellar, James Craig, Bryan Tolson, Robert Chlumsky","doi":"10.1016/j.envsoft.2026.106894","DOIUrl":"10.1016/j.envsoft.2026.106894","url":null,"abstract":"<div><div>Bayesian calibration of hydrologic models effectively addresses parameter uncertainties and improves predictions, but the joint inference of hydrologic and error model parameters often suffers from slow convergence due to high-dimensional interactions. To overcome this, a surrogate-aided error model is introduced that decouples their inference. This method uses support vector regression (SVR) as a surrogate, to estimate error model parameters conditioned on hydrologic parameters. This approach accelerates convergence (requiring 50 % fewer samples) and improves predictive accuracy, consistently improving or maintaining Continuous Ranked Probability Scores across a range of test models. These advantages are demonstrated through an application to the GR4J model across 12 MOPEX watersheds. The reduced computational demand makes this particularly valuable for large-scale hydrologic modeling when computational resources are limited.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106894"},"PeriodicalIF":4.6,"publicationDate":"2026-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048044","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}
Pub Date : 2026-01-23DOI: 10.1016/j.envsoft.2026.106891
Paul Magoulick
Coastal flooding threatens growing populations where compound hazards amplify risks. This study presents a proof-of-concept operational digital twin for single-location flood prediction at Annapolis in the Chesapeake Bay, integrating real-time NOAA and USGS data. An ensemble of Random Forest, XGBoost, Gradient Boosting, and LSTM models achieves RMSE = 0.043 ft with R = 0.997 for short-term predictions, validated across 508,000 records and 369 extreme events over six years. Strong short-term accuracy largely reflects tidal autocorrelation in this semi-enclosed estuarine system. Feature importance shows 98.9% of predictive power derives from three water-level persistence variables, enabling efficient deployment. An empirical correction factor (0.87) calibrates predictions to local conditions. Key limitations include single-site validation without spatial inundation capability and no major hurricane landfall during the study period. The system complements physics-based models such as NOAA’s STOFS, which provide essential spatial detail and process understanding. The open-source implementation enables replication and community evaluation.
{"title":"Operational digital twin for multi-hazard coastal flood prediction with adaptive learning: Real-time performance in the Chesapeake Bay","authors":"Paul Magoulick","doi":"10.1016/j.envsoft.2026.106891","DOIUrl":"10.1016/j.envsoft.2026.106891","url":null,"abstract":"<div><div>Coastal flooding threatens growing populations where compound hazards amplify risks. This study presents a proof-of-concept operational digital twin for single-location flood prediction at Annapolis in the Chesapeake Bay, integrating real-time NOAA and USGS data. An ensemble of Random Forest, XGBoost, Gradient Boosting, and LSTM models achieves RMSE = 0.043 ft with R<span><math><msup><mrow></mrow><mrow><mn>2</mn></mrow></msup></math></span> = 0.997 for short-term predictions, validated across 508,000 records and 369 extreme events over six years. Strong short-term accuracy largely reflects tidal autocorrelation in this semi-enclosed estuarine system. Feature importance shows 98.9% of predictive power derives from three water-level persistence variables, enabling efficient deployment. An empirical correction factor (0.87) calibrates predictions to local conditions. Key limitations include single-site validation without spatial inundation capability and no major hurricane landfall during the study period. The system complements physics-based models such as NOAA’s STOFS, which provide essential spatial detail and process understanding. The open-source implementation enables replication and community evaluation.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106891"},"PeriodicalIF":4.6,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146033917","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}
Pub Date : 2026-01-20DOI: 10.1016/j.envsoft.2026.106882
Mingzhuang Sun , Zhili Li , Guangtao Fu , Haifeng Jia
Water quality models often suffer from performance degradation due to parameter obsolescence caused by external environmental changes. To address this, this study proposes a novel framework named Analytical Data Assimilation via Phase-space Tuning (ADAPT). Unlike traditional data assimilation methods that directly update state variables, ADAPT dynamically calibrates model parameters by establishing a robust link between parameters and water quality dynamics using Aquaformer, a Transformer-based deep learning model driven by phase-space reconstruction. The method was validated through digital-twin and real-world experiments on the Diannong River, China. Results demonstrate that ADAPT significantly outperforms the Ensemble Kalman Filter, reducing prediction errors by 36.26 % at monitored sites and 54.66 % at unmonitored sites. ADAPT exhibits superior transferability and stable error control, effectively overcoming the limitations of traditional methods in spatial generalization. This study provides a reliable, physics-informed solution for high-frequency auto-calibration in smart water management systems.
{"title":"ADAPT: A novel IoT-driven analytical data assimilation method based on phase-space tuning for long-sequence water quality forecasting","authors":"Mingzhuang Sun , Zhili Li , Guangtao Fu , Haifeng Jia","doi":"10.1016/j.envsoft.2026.106882","DOIUrl":"10.1016/j.envsoft.2026.106882","url":null,"abstract":"<div><div>Water quality models often suffer from performance degradation due to parameter obsolescence caused by external environmental changes. To address this, this study proposes a novel framework named Analytical Data Assimilation via Phase-space Tuning (ADAPT). Unlike traditional data assimilation methods that directly update state variables, ADAPT dynamically calibrates model parameters by establishing a robust link between parameters and water quality dynamics using Aquaformer, a Transformer-based deep learning model driven by phase-space reconstruction. The method was validated through digital-twin and real-world experiments on the Diannong River, China. Results demonstrate that ADAPT significantly outperforms the Ensemble Kalman Filter, reducing prediction errors by 36.26 % at monitored sites and 54.66 % at unmonitored sites. ADAPT exhibits superior transferability and stable error control, effectively overcoming the limitations of traditional methods in spatial generalization. This study provides a reliable, physics-informed solution for high-frequency auto-calibration in smart water management systems.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106882"},"PeriodicalIF":4.6,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146014634","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}
Effective spatial monitoring of carbon fluxes is crucial for implementing climate change mitigation and adaptation measures. This study develops an advanced machine learning (ML) pipeline to assess integral carbon fluxes at regional scales using Earth observation data and ground-based measurements. We aimed to address main limitations of spatial ML assessments associated with ignorance of environmental processes’ physical nature. We propose a training pipeline ensuring prediction robustness and model generalization, introducing influential features and ground truth data selection strategy. This results in a robust mapping tool with uncertainty estimations, supported by Shapley values-based feature importance analysis for interpretability and physical meaning. Our approach utilizes data from 168 FLUXNET stations, NASA POWER meteorological reanalysis, and MODIS satellite observations to train a CatBoost gradient boosting model. The model achieves of 0.76 predicting monthly NEE values with high spatial–temporal coherence, opening possibilities for comprehensive terrestrial ecosystem carbon dynamics assessments.
{"title":"Data-driven approach to robust spatio-temporal assessment of carbon fluxes using Earth observation and ground-based data","authors":"Artem Gorbarenko , Mikhail Gasanov , Elizaveta Gorbarenko , Polina Tregubova , Anna Petrovskaia , Usman Tasuev , Svetlana Illarionova , Dmitrii Shardrin , Evgeny Burnaev","doi":"10.1016/j.envsoft.2026.106881","DOIUrl":"10.1016/j.envsoft.2026.106881","url":null,"abstract":"<div><div>Effective spatial monitoring of carbon fluxes is crucial for implementing climate change mitigation and adaptation measures. This study develops an advanced machine learning (ML) pipeline to assess integral carbon fluxes at regional scales using Earth observation data and ground-based measurements. We aimed to address main limitations of spatial ML assessments associated with ignorance of environmental processes’ physical nature. We propose a training pipeline ensuring prediction robustness and model generalization, introducing influential features and ground truth data selection strategy. This results in a robust mapping tool with uncertainty estimations, supported by Shapley values-based feature importance analysis for interpretability and physical meaning. Our approach utilizes data from 168 FLUXNET stations, NASA POWER meteorological reanalysis, and MODIS satellite observations to train a CatBoost gradient boosting model. The model achieves <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> of 0.76 predicting monthly NEE values with high spatial–temporal coherence, opening possibilities for comprehensive terrestrial ecosystem carbon dynamics assessments.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106881"},"PeriodicalIF":4.6,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146014637","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}
Pub Date : 2026-01-20DOI: 10.1016/j.envsoft.2026.106880
Shaun S.H. Kim , Russell S. Crosbie , Warrick Dawes , Jai Vaze , Bill Wang , Cherry Mateo , Rebekah May , Sudeep Nair , Jahangir Alam
Basin-scale water resources models lack key physical factors such as antecedent conditions, that strongly influence transmission losses under dry conditions. This study presents the development and evaluation of two river transmission loss models for integration into river systems models: dynamic maximum alluvium as river storage (DMAARS) and DMAARS coupled with river dead storage (DMAARSDS). The models were applied to environmental flow events during the 2018/2019 drought in the northern Murray-Darling Basin and compared with a benchmark piecewise linear loss model. They provided significantly improved performance in 8 out of 12 fit metrics and more realistic estimates of environmental flow metrics. Scenario testing revealed that model choice significantly influences predictions, especially of baseline conditions and ecological benefits, e.g., peak water height, flow extent. Analyses also showed strong potential for use in long-term water resource planning. To enable adoption, the new models have been integrated into eWater Source as a community plugin.
{"title":"Improved river transmission loss modelling for environmental flow releases during droughts","authors":"Shaun S.H. Kim , Russell S. Crosbie , Warrick Dawes , Jai Vaze , Bill Wang , Cherry Mateo , Rebekah May , Sudeep Nair , Jahangir Alam","doi":"10.1016/j.envsoft.2026.106880","DOIUrl":"10.1016/j.envsoft.2026.106880","url":null,"abstract":"<div><div>Basin-scale water resources models lack key physical factors such as antecedent conditions, that strongly influence transmission losses under dry conditions. This study presents the development and evaluation of two river transmission loss models for integration into river systems models: dynamic maximum alluvium as river storage (DMAARS) and DMAARS coupled with river dead storage (DMAARSDS). The models were applied to environmental flow events during the 2018/2019 drought in the northern Murray-Darling Basin and compared with a benchmark piecewise linear loss model. They provided significantly improved performance in 8 out of 12 fit metrics and more realistic estimates of environmental flow metrics. Scenario testing revealed that model choice significantly influences predictions, especially of baseline conditions and ecological benefits, e.g., peak water height, flow extent. Analyses also showed strong potential for use in long-term water resource planning. To enable adoption, the new models have been integrated into eWater Source as a community plugin.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106880"},"PeriodicalIF":4.6,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146014636","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}
Pub Date : 2026-01-19DOI: 10.1016/j.envsoft.2026.106868
Yibo Li, Juan Li, Simin Guo, Mei Sun
This article has developed a modular framework that integrates a provincial-level computable general equilibrium (CGE) model with a 3E3S (Energy, Economy, Environment, Sustainability, Strategy, Socio-Economic Stability) assessment to quantify the policy-driven low-carbon transition in Chinas iron and steel industry. The architecture of the framework separates data, scenarios, and solvers, incorporates mechanisms such as learning-by-doing, carbon tax, and tax-rebate instruments, and supports reproducible scenario management. When applied to policy shocks, the framework generates a composite, multi-dimensional transition impact at the provincial level. The results reveal significant spatial heterogeneity, with Hebei exhibiting the largest aggregate impact across all dimensions. In a carbon-tax and global-slowdown scenario, the transition impact reaches 0.852. This study presents detailed descriptions of the model components, I/O schemas, and workflow to facilitate reuse and adaptation for other regions or sectors. This approach demonstrates how integrated economy-wide modeling and indicator analytics can guide region-specific decarbonization strategies.
{"title":"Provincial policy simulation for steel decarbonization in China: a modular CGE-3E3S coupling framework","authors":"Yibo Li, Juan Li, Simin Guo, Mei Sun","doi":"10.1016/j.envsoft.2026.106868","DOIUrl":"10.1016/j.envsoft.2026.106868","url":null,"abstract":"<div><div>This article has developed a modular framework that integrates a provincial-level computable general equilibrium (CGE) model with a 3E3S (Energy, Economy, Environment, Sustainability, Strategy, Socio-Economic Stability) assessment to quantify the policy-driven low-carbon transition in Chinas iron and steel industry. The architecture of the framework separates data, scenarios, and solvers, incorporates mechanisms such as learning-by-doing, carbon tax, and tax-rebate instruments, and supports reproducible scenario management. When applied to policy shocks, the framework generates a composite, multi-dimensional transition impact at the provincial level. The results reveal significant spatial heterogeneity, with Hebei exhibiting the largest aggregate impact across all dimensions. In a carbon-tax and global-slowdown scenario, the transition impact reaches 0.852. This study presents detailed descriptions of the model components, I/O schemas, and workflow to facilitate reuse and adaptation for other regions or sectors. This approach demonstrates how integrated economy-wide modeling and indicator analytics can guide region-specific decarbonization strategies.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106868"},"PeriodicalIF":4.6,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146000912","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}
Pub Date : 2026-01-19DOI: 10.1016/j.envsoft.2026.106879
Yiping He , Zhaocai Wang , Heqin Cheng , Weijie Ding
Streamflow prediction, as a critical component of flood prevention and water resource management, plays a vital role in safeguarding lives, property, and socio-economic stability. However, the streamflow process is influenced by complex interactions among meteorological variability, topographic conditions, and human activities, exhibiting pronounced nonlinearity, non-stationarity, and spatiotemporal heterogeneity, which pose significant challenges to accurate prediction. This study proposes a novel hybrid streamflow prediction model, NRBO-VMD-Wavelet-TCN (NVWT), which integrates multi-source data with deep learning. The model adaptively optimizes Variational Mode Decomposition (VMD) using the Newton-Raphson-based optimizer (NRBO); it is also combined with Wavelet Thresholding Denoising (WTD), effectively suppressing noise while preserving critical hydrological signatures. A multidimensional feature system incorporating lagged, periodic, and cumulative features is constructed, after which Maximum Information Coefficient (MIC)-based feature selection is applied to enhance input efficiency. The Temporal Convolutional Network (TCN), which leverages dilated convolutions and residual connections, effectively captures long- and short-term dependencies. Experimental results across nine hydrological stations in the Hanjiang River Basin demonstrate the NVWT model's superiority in short-term prediction (1–5 days), with Nash-Sutcliffe Efficiency (NSE) exceeding 0.82 at all stations and peaking at 0.982 for the downstream Xiantao Station, significantly outperforming benchmarks (e.g., Gated Recurrent Unit (GRU), Transformer). Ablation studies confirm the efficacy of the hybrid denoising module and multidimensional features, especially during flood peaks and low-flow periods. Interval prediction metrics further validate the model's ability to quantify uncertainty. The Shapley Additive Explanation (SHAP) method reveals differential contributions of upstream lagged streamflow and meteorological factors, enhancing the model's interpretability. This study provides a methodological reference for streamflow prediction in complex watersheds, which has significant practical implications for enhancing the scientific basis of flood control decision-making under extreme climatic conditions.
{"title":"Multi-scale feature fusion and uncertainty quantification in streamflow prediction: A temporal convolutional network approach with hybrid denoising","authors":"Yiping He , Zhaocai Wang , Heqin Cheng , Weijie Ding","doi":"10.1016/j.envsoft.2026.106879","DOIUrl":"10.1016/j.envsoft.2026.106879","url":null,"abstract":"<div><div>Streamflow prediction, as a critical component of flood prevention and water resource management, plays a vital role in safeguarding lives, property, and socio-economic stability. However, the streamflow process is influenced by complex interactions among meteorological variability, topographic conditions, and human activities, exhibiting pronounced nonlinearity, non-stationarity, and spatiotemporal heterogeneity, which pose significant challenges to accurate prediction. This study proposes a novel hybrid streamflow prediction model, NRBO-VMD-Wavelet-TCN (NVWT), which integrates multi-source data with deep learning. The model adaptively optimizes Variational Mode Decomposition (VMD) using the Newton-Raphson-based optimizer (NRBO); it is also combined with Wavelet Thresholding Denoising (WTD), effectively suppressing noise while preserving critical hydrological signatures. A multidimensional feature system incorporating lagged, periodic, and cumulative features is constructed, after which Maximum Information Coefficient (MIC)-based feature selection is applied to enhance input efficiency. The Temporal Convolutional Network (TCN), which leverages dilated convolutions and residual connections, effectively captures long- and short-term dependencies. Experimental results across nine hydrological stations in the Hanjiang River Basin demonstrate the NVWT model's superiority in short-term prediction (1–5 days), with Nash-Sutcliffe Efficiency (NSE) exceeding 0.82 at all stations and peaking at 0.982 for the downstream Xiantao Station, significantly outperforming benchmarks (e.g., Gated Recurrent Unit (GRU), Transformer). Ablation studies confirm the efficacy of the hybrid denoising module and multidimensional features, especially during flood peaks and low-flow periods. Interval prediction metrics further validate the model's ability to quantify uncertainty. The Shapley Additive Explanation (SHAP) method reveals differential contributions of upstream lagged streamflow and meteorological factors, enhancing the model's interpretability. This study provides a methodological reference for streamflow prediction in complex watersheds, which has significant practical implications for enhancing the scientific basis of flood control decision-making under extreme climatic conditions.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106879"},"PeriodicalIF":4.6,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146000915","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}
Pub Date : 2026-01-17DOI: 10.1016/j.envsoft.2026.106875
Yanan Dong, Fei Wang
Enhancing ecological resilience is essential for sustainable urban development under climate and urbanization pressures. This study focuses on Xi'an's main urban area, where ecological resilience remains relatively weak. Based on the conceptual framework of resistance, elasticity, and adaptability, we construct a comprehensive evaluation model, innovatively incorporating disaster adaptability factors, and assess ecological resilience from 2000 to 2023 using a differentiated weighting method. Land Use Transition Matrices and Geographic Detectors Model (GDM) are applied to analyze spatiotemporal patterns and driving forces. Results show: (1) Ecological resilience first declined and then stabilized, mainly due to large-scale farmland conversion (273.9 km2, 32.9 %) and insufficient restoration (only 13.93 km2 added). A “weak center–strong periphery” spatial pattern emerged. (2) Resistance shifted outward (Baqiao-centered), elasticity declined, and adaptability peaked in 2023 (0.088), reflecting a transition to a society–infrastructure coupling model but constrained by central overload and peripheral deficits. (3) Driving mechanisms evolved from natural dominance to multi-factor synergy, with Landscape Shape Index (LSI)-population interaction (q = 0.22447) becoming key. The framework supports data-driven ecological resilience assessment and spatial planning in China's rapidly urbanizing regions.
{"title":"Exploring the spatiotemporal evolution and driving mechanism of ecological resilience in main urban area of Xi'an based on the “Resistance-Elasticity-Adaptability” model","authors":"Yanan Dong, Fei Wang","doi":"10.1016/j.envsoft.2026.106875","DOIUrl":"10.1016/j.envsoft.2026.106875","url":null,"abstract":"<div><div>Enhancing ecological resilience is essential for sustainable urban development under climate and urbanization pressures. This study focuses on Xi'an's main urban area, where ecological resilience remains relatively weak. Based on the conceptual framework of resistance, elasticity, and adaptability, we construct a comprehensive evaluation model, innovatively incorporating disaster adaptability factors, and assess ecological resilience from 2000 to 2023 using a differentiated weighting method. Land Use Transition Matrices and Geographic Detectors Model (GDM) are applied to analyze spatiotemporal patterns and driving forces. Results show: (1) Ecological resilience first declined and then stabilized, mainly due to large-scale farmland conversion (273.9 km<sup>2</sup>, 32.9 %) and insufficient restoration (only 13.93 km<sup>2</sup> added). A “weak center–strong periphery” spatial pattern emerged. (2) Resistance shifted outward (Baqiao-centered), elasticity declined, and adaptability peaked in 2023 (0.088), reflecting a transition to a society–infrastructure coupling model but constrained by central overload and peripheral deficits. (3) Driving mechanisms evolved from natural dominance to multi-factor synergy, with Landscape Shape Index (LSI)-population interaction (q = 0.22447) becoming key. The framework supports data-driven ecological resilience assessment and spatial planning in China's rapidly urbanizing regions.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106875"},"PeriodicalIF":4.6,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995199","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}
Pub Date : 2026-01-15DOI: 10.1016/j.envsoft.2026.106878
P. Reina-Jiménez , M.J. Jiménez-Navarro , G. Asencio-Cortés , F. Martínez-Álvarez , M. Martínez-Ballesteros
Air pollution is a growing threat, especially in low- and middle-income countries, causing over 4 million premature deaths annually. Ground-level ozone is a major concern, demanding accurate and interpretable prediction systems for effective public health management. However, existing time-series forecasting methods struggle to capture both linear and nonlinear dependencies in atmospheric data. This study introduces ResSelNet, a novel Residual Selection Network that integrates masked residual connections and embedded feature selection within a unified deep learning architecture. The model dynamically determines the optimal processing depth for each feature, allowing linear relationships to bypass nonlinear transformations while capturing complex patterns when necessary. Applied to five monitoring stations across Andalusia (Spain), ResSelNet consistently outperformed state-of-the-art baselines, achieving 8%–12% lower RMSE and MAE than LSTM and Transformer models. Beyond accuracy, the framework improves interpretability and robustness, revealing the hierarchical relevance of meteorological and pollutant variables. ResSelNet therefore offers an effective and explainable solution for multi-horizon environmental time-series forecasting.
{"title":"A novel interpretable ozone forecasting approach based on deep learning with masked residual connections","authors":"P. Reina-Jiménez , M.J. Jiménez-Navarro , G. Asencio-Cortés , F. Martínez-Álvarez , M. Martínez-Ballesteros","doi":"10.1016/j.envsoft.2026.106878","DOIUrl":"10.1016/j.envsoft.2026.106878","url":null,"abstract":"<div><div>Air pollution is a growing threat, especially in low- and middle-income countries, causing over 4 million premature deaths annually. Ground-level ozone is a major concern, demanding accurate and interpretable prediction systems for effective public health management. However, existing time-series forecasting methods struggle to capture both linear and nonlinear dependencies in atmospheric data. This study introduces ResSelNet, a novel Residual Selection Network that integrates masked residual connections and embedded feature selection within a unified deep learning architecture. The model dynamically determines the optimal processing depth for each feature, allowing linear relationships to bypass nonlinear transformations while capturing complex patterns when necessary. Applied to five monitoring stations across Andalusia (Spain), ResSelNet consistently outperformed state-of-the-art baselines, achieving 8%–12% lower RMSE and MAE than LSTM and Transformer models. Beyond accuracy, the framework improves interpretability and robustness, revealing the hierarchical relevance of meteorological and pollutant variables. ResSelNet therefore offers an effective and explainable solution for multi-horizon environmental time-series forecasting.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106878"},"PeriodicalIF":4.6,"publicationDate":"2026-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995200","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}
Pub Date : 2026-01-14DOI: 10.1016/j.envsoft.2026.106872
Jiahao Zhou , Sen He , Shanjunxia Wu , Jia Zhang , Qiuhua Wang , Fei Wang
High-quality observational data capturing the complete wildfire lifecycle are essential for validating and enhancing prediction models, yet such integrated datasets remain scarce. This study presents a modelling framework based on multitemporal data acquired through UAV sensing, including high-precision LiDAR, photogrammetry, and synchronous environmental monitoring. A multi-UAV relay observation strategy was developed to continuously record sub-second wildfire propagation dynamics. We demonstrate the utility of this framework through benchmark modelling experiments in fuel mapping, fire spread prediction, and burn severity assessment. The high-resolution data provide a valuable and comprehensive basis for evaluating model behavior across temporal scales, particularly in capturing early fire progression and fire-atmosphere interactions. It also reveals limitations in current modelling approaches. This work offers a robust resource for advancing wildfire environmental modelling.
{"title":"THU-Wildfire: A multitemporal, multimodal observation dataset for wildfire behavior dynamics","authors":"Jiahao Zhou , Sen He , Shanjunxia Wu , Jia Zhang , Qiuhua Wang , Fei Wang","doi":"10.1016/j.envsoft.2026.106872","DOIUrl":"10.1016/j.envsoft.2026.106872","url":null,"abstract":"<div><div>High-quality observational data capturing the complete wildfire lifecycle are essential for validating and enhancing prediction models, yet such integrated datasets remain scarce. This study presents a modelling framework based on multitemporal data acquired through UAV sensing, including high-precision LiDAR, photogrammetry, and synchronous environmental monitoring. A multi-UAV relay observation strategy was developed to continuously record sub-second wildfire propagation dynamics. We demonstrate the utility of this framework through benchmark modelling experiments in fuel mapping, fire spread prediction, and burn severity assessment. The high-resolution data provide a valuable and comprehensive basis for evaluating model behavior across temporal scales, particularly in capturing early fire progression and fire-atmosphere interactions. It also reveals limitations in current modelling approaches. This work offers a robust resource for advancing wildfire environmental modelling.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"198 ","pages":"Article 106872"},"PeriodicalIF":4.6,"publicationDate":"2026-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995201","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}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号