Pub Date : 2025-11-19DOI: 10.1016/j.envsoft.2025.106771
Daniel Velez-Serrano , Alejandro Alvaro-Meca
This study presents a novel deep learning-based model, the Improved Spatio-Temporal Graph Transformer (ISTGT), designed for accurate municipal-level PM2.5 forecasting across Spain. ISTGT integrates Graph Convolutional Networks, Temporal Convolutional Networks, and Transformer Encoders to capture complex spatial relationships and temporal dependencies. An adaptive spatial graph, constructed using Delaunay triangulation, incorporates distance, altitude, and population density to enhance prediction accuracy. Historical data — including air quality, meteorological factors, elevation, population, and public holidays — from 8,076 municipalities facilitated detailed predictions and extrapolation onto a fine-resolution spatial grid (0.1° × 0.1°). Combining ISTGT with ARIMA predictions using a CatBoost stacking approach significantly reduced mean absolute error (MAE) to 1.24, outperforming traditional and hybrid models. The proposed method offers computational efficiency, precise spatial extrapolation, and adaptability to other spatio-temporal tasks, providing a valuable tool for environmental management. Future work may integrate real-time meteorological and satellite data to improve predictions during extreme conditions.
{"title":"Adaptive Graph Neural Network–transformer model for high-resolution PM2.5 forecasting and spatial extrapolation","authors":"Daniel Velez-Serrano , Alejandro Alvaro-Meca","doi":"10.1016/j.envsoft.2025.106771","DOIUrl":"10.1016/j.envsoft.2025.106771","url":null,"abstract":"<div><div>This study presents a novel deep learning-based model, the Improved Spatio-Temporal Graph Transformer (ISTGT), designed for accurate municipal-level PM<sub>2.5</sub> forecasting across Spain. ISTGT integrates Graph Convolutional Networks, Temporal Convolutional Networks, and Transformer Encoders to capture complex spatial relationships and temporal dependencies. An adaptive spatial graph, constructed using Delaunay triangulation, incorporates distance, altitude, and population density to enhance prediction accuracy. Historical data — including air quality, meteorological factors, elevation, population, and public holidays — from 8,076 municipalities facilitated detailed predictions and extrapolation onto a fine-resolution spatial grid (0.1° × 0.1°). Combining ISTGT with ARIMA predictions using a CatBoost stacking approach significantly reduced mean absolute error (MAE) to 1.24, outperforming traditional and hybrid models. The proposed method offers computational efficiency, precise spatial extrapolation, and adaptability to other spatio-temporal tasks, providing a valuable tool for environmental management. Future work may integrate real-time meteorological and satellite data to improve predictions during extreme conditions.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106771"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560062","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 : 2025-11-19DOI: 10.1016/j.envsoft.2025.106791
Fan Liu, Zhao Guo, Chen Ma, Futian Ren, Zenghui Li, Xiaowei Lu, Lei Huang
An AI-enhanced, cloud-native platform for groundwater management integrates physics-based simulation, data-driven surrogates, and Bayesian uncertainty quantification. The framework couples MODFLOW-6 with a Random-Forest (RF) surrogate and a prototype Physics-Informed Neural Network (PINN), supporting ensemble calibration (PyEMU) and surrogate-driven probabilistic inference. In an industrial-park application, the calibrated MF6 reproduced observed heads (RMSE 0.40; MAE 0.32; NSE 0.84). The RF surrogate maintained high fidelity (validation NSE 0.78) with reduced computational cost, while the PINN enforced physical constraints but showed lower pointwise accuracy. Both inference methods identified hydraulic conductivity as the dominant sensitive parameter and provided credible intervals and exceedance probabilities for risk assessment. A web interface enables data ingestion, model setup, scenario exploration, and uncertainty-aware visualization, including 3D flow/plume, residual maps, and time-series warnings. This platform offers a reproducible, scalable, and physically consistent pathway for operational groundwater decision support and future enhancements such as neural operators and reactive transport modeling.
人工智能增强的地下水管理云原生平台集成了基于物理的模拟、数据驱动的替代和贝叶斯不确定性量化。该框架将MODFLOW-6与随机森林（RF）代理和原型物理信息神经网络（PINN）耦合在一起，支持集成校准（PyEMU）和代理驱动的概率推理。在一个工业园区的应用中，校准后的MF6再现了观察到的头部（RMSE 0.40; MAE 0.32; NSE 0.84）。RF代理保持了高保真度（验证NSE 0.78）并降低了计算成本，而PINN强制物理约束但显示出较低的点精度。两种推理方法均将导电性作为主导敏感参数，并为风险评估提供可信区间和超出概率。web界面支持数据摄取、模型设置、场景探索和不确定性感知可视化，包括3D流/羽流、残余地图和时间序列警告。该平台为地下水作业决策支持和未来的增强功能（如神经算子和反应性输运建模）提供了可复制、可扩展和物理一致的途径。
{"title":"AI-enhanced groundwater management platform: A network-driven approach for simulation","authors":"Fan Liu, Zhao Guo, Chen Ma, Futian Ren, Zenghui Li, Xiaowei Lu, Lei Huang","doi":"10.1016/j.envsoft.2025.106791","DOIUrl":"10.1016/j.envsoft.2025.106791","url":null,"abstract":"<div><div>An AI-enhanced, cloud-native platform for groundwater management integrates physics-based simulation, data-driven surrogates, and Bayesian uncertainty quantification. The framework couples MODFLOW-6 with a Random-Forest (RF) surrogate and a prototype Physics-Informed Neural Network (PINN), supporting ensemble calibration (PyEMU) and surrogate-driven probabilistic inference. In an industrial-park application, the calibrated MF6 reproduced observed heads (RMSE 0.40; MAE 0.32; NSE 0.84). The RF surrogate maintained high fidelity (validation NSE 0.78) with reduced computational cost, while the PINN enforced physical constraints but showed lower pointwise accuracy. Both inference methods identified hydraulic conductivity as the dominant sensitive parameter and provided credible intervals and exceedance probabilities for risk assessment. A web interface enables data ingestion, model setup, scenario exploration, and uncertainty-aware visualization, including 3D flow/plume, residual maps, and time-series warnings. This platform offers a reproducible, scalable, and physically consistent pathway for operational groundwater decision support and future enhancements such as neural operators and reactive transport modeling.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106791"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553967","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 : 2025-11-19DOI: 10.1016/j.envsoft.2025.106795
Huili Wang , Bin Xu , Xinman Qin , Xinrong Wang , Jianyun Zhang , Guoqing Wang , Fubao Yang , Ping-an Zhong , Ran Mo , Xuesong Yang
Traditional methods for simulating reservoir scheduling rule face challenges in reducing spatiotemporal errors and improving Pareto frontier simulation quality for multi-objective optimization. This study proposes a spatiotemporal correction of decision variables technique using XGBoost (SC-XGB) to extract intelligent multi-objective scheduling rules. A two-stage scheduling rule framework is designed to reduce model complexity, and a spatiotemporal correction loss function is introduced to mitigate cumulative water balance constraint violation errors. Bayesian optimization with cross-validation is employed for hyperparameter tuning, and a multi-metric evaluation system is established. Case study results from the Chaohu Basin, China, show that the SC-XGB improves the average NSE of outflow prediction by 1.89 %, reduces the Water Balance Mean Error range of Chaohu Lake by 27.93 %, and decreases the Relative Hypervolume Error by 21.51 % compared to the XGB model. These findings demonstrate that the SC-XGB model enhances both accuracy and generalization, thereby supporting intelligent scheduling in flood management systems.
{"title":"Spatiotemporal correction of decision variables using XGBoost for multi-objective intelligent scheduling rule extraction model in reservoir-lake flood control systems","authors":"Huili Wang , Bin Xu , Xinman Qin , Xinrong Wang , Jianyun Zhang , Guoqing Wang , Fubao Yang , Ping-an Zhong , Ran Mo , Xuesong Yang","doi":"10.1016/j.envsoft.2025.106795","DOIUrl":"10.1016/j.envsoft.2025.106795","url":null,"abstract":"<div><div>Traditional methods for simulating reservoir scheduling rule face challenges in reducing spatiotemporal errors and improving Pareto frontier simulation quality for multi-objective optimization. This study proposes a spatiotemporal correction of decision variables technique using XGBoost (SC-XGB) to extract intelligent multi-objective scheduling rules. A two-stage scheduling rule framework is designed to reduce model complexity, and a spatiotemporal correction loss function is introduced to mitigate cumulative water balance constraint violation errors. Bayesian optimization with cross-validation is employed for hyperparameter tuning, and a multi-metric evaluation system is established. Case study results from the Chaohu Basin, China, show that the SC-XGB improves the average NSE of outflow prediction by 1.89 %, reduces the Water Balance Mean Error range of Chaohu Lake by 27.93 %, and decreases the Relative Hypervolume Error by 21.51 % compared to the XGB model. These findings demonstrate that the SC-XGB model enhances both accuracy and generalization, thereby supporting intelligent scheduling in flood management systems.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106795"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145560061","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 : 2025-11-19DOI: 10.1016/j.envsoft.2025.106790
Chengxin Qin , Fu Sun , Yi Rong , Wanbin Wang , Xingzi Zhang , Yihui Chen , Yi Liu
Model evaluation is crucial for verifying model credibility, especially in decision-making. Successful environmental modelling requires not only self-proved credibility from model developers/users and peer-appraised credibility from technical experts, but also decision-maker and public confidence in model credibility. We propose a participatory model evaluation approach for environmental decisions, combining the standard evaluation procedure, data-augmented peer review and multi-stakeholder engagement. To facilitate this approach, we developed DPMODE (Decision Procedure Management of surface water mODel Evaluation), a web-based system with supporting tools and database. DPMODE evaluates surface water models and recommends credible models and customized test datasets for watershed management. A case study on the Soil and Water Assessment Tool (SWAT) for the Chishui River watershed management demonstrated the effectiveness of this approach. This participatory evaluation would be an adaptive, iterative process to improve stakeholder acceptance, enhance model-based outcomes, and foster better decision pathways.
{"title":"Developing a web-based participatory approach to model evaluation for environmental decision-making","authors":"Chengxin Qin , Fu Sun , Yi Rong , Wanbin Wang , Xingzi Zhang , Yihui Chen , Yi Liu","doi":"10.1016/j.envsoft.2025.106790","DOIUrl":"10.1016/j.envsoft.2025.106790","url":null,"abstract":"<div><div>Model evaluation is crucial for verifying model credibility, especially in decision-making. Successful environmental modelling requires not only self-proved credibility from model developers/users and peer-appraised credibility from technical experts, but also decision-maker and public confidence in model credibility. We propose a participatory model evaluation approach for environmental decisions, combining the standard evaluation procedure, data-augmented peer review and multi-stakeholder engagement. To facilitate this approach, we developed DPMODE (Decision Procedure Management of surface water mODel Evaluation), a web-based system with supporting tools and database. DPMODE evaluates surface water models and recommends credible models and customized test datasets for watershed management. A case study on the Soil and Water Assessment Tool (SWAT) for the Chishui River watershed management demonstrated the effectiveness of this approach. This participatory evaluation would be an adaptive, iterative process to improve stakeholder acceptance, enhance model-based outcomes, and foster better decision pathways.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106790"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553968","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 : 2025-11-19DOI: 10.1016/j.envsoft.2025.106788
Fletcher T. Chapin , Yin-Li Liu , Meagan S. Mauter
Digital twins and other digital solutions are transforming the planning, design, operation, and maintenance of water assets. Implementing these solutions is often slowed by data management activities including cleaning, storage, and querying. We identify three limitations of existing data management platforms: data inaccessibility, inadequate integration of data and metadata, and the absence of embedded data analysis capabilities. We introduce Python for Process Engineering Schema (PyPES), an object-oriented, open-source schema for water data management, to address these shortcomings. Next, we demonstrate PyPES implementation across three distinct water asset classes (water distribution, reverse osmosis, and wastewater treatment) and applications (leakage detection, optimal sensor placement, and automated fault detection). In each case study, we highlight how novel features of PyPES increase the value and portability of these models relative to state-of-the-art approaches. Finally, we describe opportunities for integrating PyPES with a data ontology to enhance the power of this software.
数字孪生和其他数字解决方案正在改变水资产的规划、设计、运营和维护。数据管理活动（包括清理、存储和查询）通常会减慢实现这些解决方案的速度。我们确定了现有数据管理平台的三个局限性：数据不可访问、数据和元数据集成不足以及缺乏嵌入式数据分析功能。我们介绍了Python for Process Engineering Schema (PyPES)，这是一种面向对象的、用于水数据管理的开源模式，以解决这些缺点。接下来，我们将演示在三种不同的水资产类别（配水、反渗透和废水处理）和应用（泄漏检测、最佳传感器放置和自动故障检测）中实现PyPES。在每个案例研究中，我们强调了PyPES的新特性如何提高这些模型相对于最先进方法的价值和可移植性。最后，我们描述了将PyPES与数据本体集成以增强该软件功能的机会。
{"title":"PyPES: A data and metadata schema for portable water system models","authors":"Fletcher T. Chapin , Yin-Li Liu , Meagan S. Mauter","doi":"10.1016/j.envsoft.2025.106788","DOIUrl":"10.1016/j.envsoft.2025.106788","url":null,"abstract":"<div><div>Digital twins and other digital solutions are transforming the planning, design, operation, and maintenance of water assets. Implementing these solutions is often slowed by data management activities including cleaning, storage, and querying. We identify three limitations of existing data management platforms: data inaccessibility, inadequate integration of data and metadata, and the absence of embedded data analysis capabilities. We introduce Python for Process Engineering Schema (PyPES), an object-oriented, open-source schema for water data management, to address these shortcomings. Next, we demonstrate PyPES implementation across three distinct water asset classes (water distribution, reverse osmosis, and wastewater treatment) and applications (leakage detection, optimal sensor placement, and automated fault detection). In each case study, we highlight how novel features of PyPES increase the value and portability of these models relative to state-of-the-art approaches. Finally, we describe opportunities for integrating PyPES with a data ontology to enhance the power of this software.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106788"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553963","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 : 2025-11-19DOI: 10.1016/j.envsoft.2025.106794
Mikołaj Piniewski , Natalja Čerkasova , Svajunas Plunge , Michael Strauch , Christoph Schürz , Péter Braun , Enrico Antonio Chiaradia , Joana Eichenberger , Mohammad Reza Eini , Csilla Farkas , Marie Anne Eurie Forio , Peter Goethals , Piroska Kassai , Štěpán Marval , Diego G. Panique-Casso , Lorenzo Sanguanini , Moritz Shore , Brigitta Szabó , Petr Slavík , Felix Witing
This study proposes a new workflow for crop growth evaluation and yield calibration in the Soil and Water Assessment Tool Plus (SWAT+) model and evaluates its impact on simulated hydrological and biogeochemical processes. The workflow was applied for ten small agricultural catchments in Europe. A detailed demonstration is provided for the German catchment, Schwarzer Schöps. The workflow proved effective across all catchments, improving yield calibration from an initial R2 of 0.5–0.84. The results show that evapotranspiration and soil moisture were only moderately affected by crop calibration in three catchments (Belgium, Czech Republic and Norway) and negligibly changed in the remaining ones. Sediment and nutrient balance were affected more strongly: sediment, nitrogen and phosphorus loss change reached 82 % (Norway), 16 % and 20 % (Czech Republic), respectively. The proposed workflow is a valuable tool for improving the accuracy of SWAT + simulations and can be used to support decision-making in environmental management.
{"title":"Enhanced crop calibration for SWAT+: evaluating water, sediment and nutrient impacts across ten European catchments","authors":"Mikołaj Piniewski , Natalja Čerkasova , Svajunas Plunge , Michael Strauch , Christoph Schürz , Péter Braun , Enrico Antonio Chiaradia , Joana Eichenberger , Mohammad Reza Eini , Csilla Farkas , Marie Anne Eurie Forio , Peter Goethals , Piroska Kassai , Štěpán Marval , Diego G. Panique-Casso , Lorenzo Sanguanini , Moritz Shore , Brigitta Szabó , Petr Slavík , Felix Witing","doi":"10.1016/j.envsoft.2025.106794","DOIUrl":"10.1016/j.envsoft.2025.106794","url":null,"abstract":"<div><div>This study proposes a new workflow for crop growth evaluation and yield calibration in the Soil and Water Assessment Tool Plus (SWAT+) model and evaluates its impact on simulated hydrological and biogeochemical processes. The workflow was applied for ten small agricultural catchments in Europe. A detailed demonstration is provided for the German catchment, Schwarzer Schöps. The workflow proved effective across all catchments, improving yield calibration from an initial R<sup>2</sup> of 0.5–0.84. The results show that evapotranspiration and soil moisture were only moderately affected by crop calibration in three catchments (Belgium, Czech Republic and Norway) and negligibly changed in the remaining ones. Sediment and nutrient balance were affected more strongly: sediment, nitrogen and phosphorus loss change reached 82 % (Norway), 16 % and 20 % (Czech Republic), respectively. The proposed workflow is a valuable tool for improving the accuracy of SWAT + simulations and can be used to support decision-making in environmental management.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106794"},"PeriodicalIF":4.6,"publicationDate":"2025-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553964","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 : 2025-11-18DOI: 10.1016/j.envsoft.2025.106787
Yan Zhu , Zhi Dou , Chaoqi Wang , Meng Chen , Yun Yang , Jinguo Wang
Groundwater contamination source identification (GCSI) is critical for water resources management but depends on the accurate characterization of aquifer parameters, especially hydraulic conductivity (K). A novel multimodal direct forward machine learning (MDFML) model was developed to simultaneously predict GCSI parameters and reconstruct K-fields. This model utilizes constrained residual fusion to integrate temporal concentration and spatial head data, and improve complementarity. Tested on synthetic Gaussian and non-Gaussian aquifers, MDFML consistently outperformed single-modal models. In Gaussian fields, MDFML improved source parameter prediction by 2.20 % (R2) and K-field reconstruction by 7.50 % (SSIM, structural similarity index) compared to single-modal baselines. In non-Gaussian fields, structured dispersion patterns achieved higher K-field reconstruction (SSIM = 0.951, +6.70 % vs. 0.892 for Gaussian), but nonlinearity lowered source prediction accuracy (R2 = 0.900, −2.75 % vs. 0.925 for Gaussian). These results demonstrate the robustness and reliability of MDFML under complex hydrogeological conditions and provide an efficient solution for accurate GCSI and sustainable groundwater remediation.
{"title":"Simultaneous identification of a contamination source and hydraulic conductivity based on a multimodal direct forward machine learning model","authors":"Yan Zhu , Zhi Dou , Chaoqi Wang , Meng Chen , Yun Yang , Jinguo Wang","doi":"10.1016/j.envsoft.2025.106787","DOIUrl":"10.1016/j.envsoft.2025.106787","url":null,"abstract":"<div><div>Groundwater contamination source identification (GCSI) is critical for water resources management but depends on the accurate characterization of aquifer parameters, especially hydraulic conductivity (K). A novel multimodal direct forward machine learning (MDFML) model was developed to simultaneously predict GCSI parameters and reconstruct K-fields. This model utilizes constrained residual fusion to integrate temporal concentration and spatial head data, and improve complementarity. Tested on synthetic Gaussian and non-Gaussian aquifers, MDFML consistently outperformed single-modal models. In Gaussian fields, MDFML improved source parameter prediction by 2.20 % (<em>R</em><sup>2</sup>) and K-field reconstruction by 7.50 % (SSIM, structural similarity index) compared to single-modal baselines. In non-Gaussian fields, structured dispersion patterns achieved higher K-field reconstruction (SSIM = 0.951, +6.70 % vs. 0.892 for Gaussian), but nonlinearity lowered source prediction accuracy (<em>R</em><sup>2</sup> = 0.900, −2.75 % vs. 0.925 for Gaussian). These results demonstrate the robustness and reliability of MDFML under complex hydrogeological conditions and provide an efficient solution for accurate GCSI and sustainable groundwater remediation.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106787"},"PeriodicalIF":4.6,"publicationDate":"2025-11-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553965","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}
Accurate Flood Inundation Mapping (FIM) is essential for forecasting and evaluation. Traditional pixel-based approaches can be time-intensive and error-prone. Here, we introduced the Flood Inundation Mapping Evaluation Framework (FIMeval), an open-source toolset for large-scale FIM evaluation. FIMeval links to a benchmarking database that includes high-quality FIM benchmarks across the Contiguous United States, derived from remote sensing and high-fidelity model-predicted datasets. FIMeval supports pixel-based metrics and integrates impact-based assessments using building footprint data. We demonstrated its application using (a) high-resolution aerial imagery FIM for 2016 Midwest Flood (b) remote sensing-derived benchmarks from Hurricane Matthew (2016), and (b) simulated 100-year and 500-year FIM across 45 Hydrologic Unit Code-8 watersheds using the Federal Emergency Management Agency's Base Level Engineering dataset. The NOAA Office of Water Prediction Height Above Nearest Drainage (OWP HAND-FIM) was the model-predicted FIM for all case studies. We tested the influence of data-imbalance on the scores using two inbuilt methods.
{"title":"A framework for the evaluation of flood inundation predictions over extensive benchmark databases","authors":"Dipsikha Devi , Supath Dhital , Dinuke Munasinghe , Sagy Cohen , Anupal Baruah , Yixian Chen , Dan Tian , Carson Pruitt","doi":"10.1016/j.envsoft.2025.106786","DOIUrl":"10.1016/j.envsoft.2025.106786","url":null,"abstract":"<div><div>Accurate Flood Inundation Mapping (FIM) is essential for forecasting and evaluation. Traditional pixel-based approaches can be time-intensive and error-prone. Here, we introduced the Flood Inundation Mapping Evaluation Framework (FIMeval), an open-source toolset for large-scale FIM evaluation. FIMeval links to a benchmarking database that includes high-quality FIM benchmarks across the Contiguous United States, derived from remote sensing and high-fidelity model-predicted datasets. FIMeval supports pixel-based metrics and integrates impact-based assessments using building footprint data. We demonstrated its application using (a) high-resolution aerial imagery FIM for 2016 Midwest Flood (b) remote sensing-derived benchmarks from Hurricane Matthew (2016), and (b) simulated 100-year and 500-year FIM across 45 Hydrologic Unit Code-8 watersheds using the Federal Emergency Management Agency's Base Level Engineering dataset. The NOAA Office of Water Prediction Height Above Nearest Drainage (OWP HAND-FIM) was the model-predicted FIM for all case studies. We tested the influence of data-imbalance on the scores using two inbuilt methods.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106786"},"PeriodicalIF":4.6,"publicationDate":"2025-11-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145553966","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 : 2025-11-16DOI: 10.1016/j.envsoft.2025.106784
Xianfeng Teng , Fangjie Mao , Huaqiang Du , Xuejian Li , Fengfeng Ye , Zhaodong Zheng , Ningxin Yang , Yinyin Zhao , Jiacong Yu , Meixuan Song
Subtropical coniferous forests serve as vital carbon sinks, with their net ecosystem productivity (NEP) significantly influenced by extreme climate events. This study enhances the BIOME-BGC model's performance by implementing dynamic heat and water stress mechanisms. Validation against 2003–2010 observational data shows substantial improvements, with correlation increasing by 51.32 % and root mean square error decreasing by 15.16 %. Analysis of NEP patterns (1981–2019) reveals an increase from 92.28 to 129.55 gC·m−2·yr−1 between the periods 1981–2000 and 2001–2019, particularly in eastern subtropical regions. Extreme drought events account for 79.45 % of low-NEP years, while extreme heat positively affects NEP in 42.97 % of high-altitude western areas. The model demonstrates enhanced sensitivity to extreme climate events, with drought showing the strongest negative impact (sensitivity: 0.43) and wet conditions promoting NEP in 63.71 % of the study area. These improvements provide robust tools for forest management and carbon dynamics assessment under changing climatic conditions.
{"title":"Incorporating heat and water stress into BIOME-BGC to simulate the impact of extreme climate events on subtropical coniferous forest NEP","authors":"Xianfeng Teng , Fangjie Mao , Huaqiang Du , Xuejian Li , Fengfeng Ye , Zhaodong Zheng , Ningxin Yang , Yinyin Zhao , Jiacong Yu , Meixuan Song","doi":"10.1016/j.envsoft.2025.106784","DOIUrl":"10.1016/j.envsoft.2025.106784","url":null,"abstract":"<div><div>Subtropical coniferous forests serve as vital carbon sinks, with their net ecosystem productivity (NEP) significantly influenced by extreme climate events. This study enhances the BIOME-BGC model's performance by implementing dynamic heat and water stress mechanisms. Validation against 2003–2010 observational data shows substantial improvements, with correlation increasing by 51.32 % and root mean square error decreasing by 15.16 %. Analysis of NEP patterns (1981–2019) reveals an increase from 92.28 to 129.55 gC·m<sup>−2</sup>·yr<sup>−1</sup> between the periods 1981–2000 and 2001–2019, particularly in eastern subtropical regions. Extreme drought events account for 79.45 % of low-NEP years, while extreme heat positively affects NEP in 42.97 % of high-altitude western areas. The model demonstrates enhanced sensitivity to extreme climate events, with drought showing the strongest negative impact (sensitivity: 0.43) and wet conditions promoting NEP in 63.71 % of the study area. These improvements provide robust tools for forest management and carbon dynamics assessment under changing climatic conditions.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106784"},"PeriodicalIF":4.6,"publicationDate":"2025-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145531089","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 : 2025-11-14DOI: 10.1016/j.envsoft.2025.106785
Babak Masoudi , Safoora Bazzi
Dust storms in the Sistan Basin, a transboundary hotspot between Iran and Afghanistan, impact public health and stability. While deep learning can forecast these events, attributing their drivers is a key challenge. We propose a Dual-Stream Graph Neural Network (DSC-GNN) for both forecasting and driver attribution. Trained on a six-year MERRA-2 dataset, our model achieves high predictive performance (R2 = 0.761) and outperforms standard baselines. A robust attribution analysis was then applied to the 50 most severe storm events. Results reveal a complex interplay between drivers: on average, atmospheric forcing contributed ∼58 % to dust intensity, while ground conditions contributed ∼42 %. Critically, the high variance in these contributions across events, a key finding supported by HYSPLIT analysis, indicates that a singular mitigation approach is insufficient. Our work suggests that effective dust management in Sistan requires a dual-pronged strategy addressing both local land rehabilitation and regional cooperation on transboundary sources.
{"title":"A dual-stream spatio-temporal graph neural network for dust storm forecasting and attribution in the Sistan Basin","authors":"Babak Masoudi , Safoora Bazzi","doi":"10.1016/j.envsoft.2025.106785","DOIUrl":"10.1016/j.envsoft.2025.106785","url":null,"abstract":"<div><div>Dust storms in the Sistan Basin, a transboundary hotspot between Iran and Afghanistan, impact public health and stability. While deep learning can forecast these events, attributing their drivers is a key challenge. We propose a Dual-Stream Graph Neural Network (DSC-GNN) for both forecasting and driver attribution. Trained on a six-year MERRA-2 dataset, our model achieves high predictive performance (R<sup>2</sup> = 0.761) and outperforms standard baselines. A robust attribution analysis was then applied to the 50 most severe storm events. Results reveal a complex interplay between drivers: on average, atmospheric forcing contributed ∼58 % to dust intensity, while ground conditions contributed ∼42 %. Critically, the high variance in these contributions across events, a key finding supported by HYSPLIT analysis, indicates that a singular mitigation approach is insufficient. Our work suggests that effective dust management in Sistan requires a dual-pronged strategy addressing both local land rehabilitation and regional cooperation on transboundary sources.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106785"},"PeriodicalIF":4.6,"publicationDate":"2025-11-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145531126","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