Pub Date : 2025-12-05DOI: 10.1016/j.envsoft.2025.106818
Xuesong Yang , Bin Xu , Huili Wang , Xinman Qin , Xinrong Wang , Zichen Ren , Yao Yao , Siying Zhou , Yao Liu , Ping Chang
Complex flood control systems which comprise reservoirs, lakes, and external rivers, frequently encounter multifaceted risk sources that are spatiotemporally interconnected, resulting in diverse flood risks. This study developed a comprehensive risk analysis framework integrating stochastic simulation and Bayesian networks to facilitate refined risk prediction and diagnosis. Vine copula and Monte Carlo methods were used for probabilistic modeling and simulation, while Bayesian network was used for bidirectional risk assessment. A case study of Chaohu Lake Basin (China) show that vine copula effectively elucidates both intervariable correlations and single variable characteristics. The lateral inflow volume of lake and the external river water levels are dominant risk sources. When the maximum water level of lake increases from 9.5 m to 11.5 m, the posterior probability of dominant risk sources exceeding the design value at 20 % increases by 46.12 % and 32.22 %. This study represents an innovative approach to risk analysis for complex reservoir-lake systems.
{"title":"Hybrid high-dimensional vine copula–Bayesian network framework for flood risk analysis in reservoir–lake systems: Addressing multisource uncertainties","authors":"Xuesong Yang , Bin Xu , Huili Wang , Xinman Qin , Xinrong Wang , Zichen Ren , Yao Yao , Siying Zhou , Yao Liu , Ping Chang","doi":"10.1016/j.envsoft.2025.106818","DOIUrl":"10.1016/j.envsoft.2025.106818","url":null,"abstract":"<div><div>Complex flood control systems which comprise reservoirs, lakes, and external rivers, frequently encounter multifaceted risk sources that are spatiotemporally interconnected, resulting in diverse flood risks. This study developed a comprehensive risk analysis framework integrating stochastic simulation and Bayesian networks to facilitate refined risk prediction and diagnosis. Vine copula and Monte Carlo methods were used for probabilistic modeling and simulation, while Bayesian network was used for bidirectional risk assessment. A case study of Chaohu Lake Basin (China) show that vine copula effectively elucidates both intervariable correlations and single variable characteristics. The lateral inflow volume of lake and the external river water levels are dominant risk sources. When the maximum water level of lake increases from 9.5 m to 11.5 m, the posterior probability of dominant risk sources exceeding the design value at 20 % increases by 46.12 % and 32.22 %. This study represents an innovative approach to risk analysis for complex reservoir-lake systems.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106818"},"PeriodicalIF":4.6,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145689374","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 : 2025-12-05DOI: 10.1016/j.envsoft.2025.106824
Ali Azedou , Aouatif Amine , Said Lahssini , Gordon Osterman , Mauricio Arboleda-Zapata , Michael Cosh , Isaya Kisekka
Soil moisture (SM) plays a vital role in both hydrological and agricultural processes and is critical for achieving groundwater sustainability in agriculture through demand management. NASA's Soil Moisture Active Passive (SMAP) satellite measures the SM across the Earth and provides data on both the surface and root zone SM but at a coarse spatial resolution of 9 km, thereby limiting detailed analyses. This study aimed to develop an optimized deep ensemble learning framework to downscale the resolution of SMAP observations of California's Central Valley from 9 km to 30 m for both the surface and root-zone SM. Sensitivity analysis was employed to identify key explanatory variables. The models were then combined into an ensemble DNN trained on multiscale SMAP data and validated against in-situ SM measurements. The results demonstrated that the ensemble model achieved the highest coefficients of determination () of 0.789 and 0.683 for surface and root-zone SM, respectively, with the lowest root mean square errors of 0.0281 and 0.0814 cm3/cm3, respectively, along with the highest NSE scores of 0.50 and 0.433, thereby reliably capturing spatial patterns and predictive accuracy. A sensitivity analysis identified precipitation, LST, topographic factors, land cover, and vegetation indices as key predictors for SSM, while organic matter, silt content, precipitation, and DEM were the most influential for RZSM. Seasonal analysis revealed distinct patterns linked to climate and management practices at a spatial resolution of 30 m, thereby capturing seasonal variations in soil moisture among major crops. Additionally, SM maps can be used to refine the estimated evapotranspiration resulting from applied irrigation water sourced from groundwater pumping, allowing for better monitoring of water use. SM can also be used to inform agronomic practices, such as delayed irrigation in early spring, which can reduce groundwater demand.
{"title":"Ensemble deep learning towards high-resolution soil-moisture mapping for enhanced water management in California's Central Valley","authors":"Ali Azedou , Aouatif Amine , Said Lahssini , Gordon Osterman , Mauricio Arboleda-Zapata , Michael Cosh , Isaya Kisekka","doi":"10.1016/j.envsoft.2025.106824","DOIUrl":"10.1016/j.envsoft.2025.106824","url":null,"abstract":"<div><div>Soil moisture (SM) plays a vital role in both hydrological and agricultural processes and is critical for achieving groundwater sustainability in agriculture through demand management. NASA's Soil Moisture Active Passive (SMAP) satellite measures the SM across the Earth and provides data on both the surface and root zone SM but at a coarse spatial resolution of 9 km, thereby limiting detailed analyses. This study aimed to develop an optimized deep ensemble learning framework to downscale the resolution of SMAP observations of California's Central Valley from 9 km to 30 m for both the surface and root-zone SM. Sensitivity analysis was employed to identify key explanatory variables. The models were then combined into an ensemble DNN trained on multiscale SMAP data and validated against in-situ SM measurements. The results demonstrated that the ensemble model achieved the highest coefficients of determination (<span><math><mrow><msup><mi>R</mi><mn>2</mn></msup></mrow></math></span>) of 0.789 and 0.683 for surface and root-zone SM, respectively, with the lowest root mean square errors of 0.0281 and 0.0814 cm3/cm3, respectively, along with the highest NSE scores of 0.50 and 0.433, thereby reliably capturing spatial patterns and predictive accuracy. A sensitivity analysis identified precipitation, LST, topographic factors, land cover, and vegetation indices as key predictors for SSM, while organic matter, silt content, precipitation, and DEM were the most influential for RZSM. Seasonal analysis revealed distinct patterns linked to climate and management practices at a spatial resolution of 30 m, thereby capturing seasonal variations in soil moisture among major crops. Additionally, SM maps can be used to refine the estimated evapotranspiration resulting from applied irrigation water sourced from groundwater pumping, allowing for better monitoring of water use. SM can also be used to inform agronomic practices, such as delayed irrigation in early spring, which can reduce groundwater demand.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106824"},"PeriodicalIF":4.6,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145689339","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 : 2025-12-05DOI: 10.1016/j.envsoft.2025.106822
Sofia Bosi , Alessandra Raffaetà , Marta Simeoni , Nikola Bobchev , Dimitar Berov , Andrea Barbanti , Stefano Menegon
PMAR (Pressure models for MARine activities) is a modelling framework and open-source Python-based software designed to assess anthropogenic pressures for marine management. PMAR uses Lagrangian trajectories calculated from ocean models to simulate pressure dispersion. An explicit link between pressures and their sources is established through a weight-based mechanism, which allows to rapidly explore different pressure source scenarios. Here, PMAR is adopted to investigate the distribution of surface macroplastics in the Black Sea under two scenarios, yielding results consistent with previous modelling studies and observations. Over 50% of the macroplastics in each Black Sea country after a yearly run came from elsewhere, stressing the importance of cross-boundary cooperation. Compared to other pressure modelling frameworks, PMAR emerges as a balanced compromise between computational efficiency and predictive accuracy. Future work will focus on making PMAR accessible to decision-makers and aligning it with requirements of Digital Twin of the Ocean applications.
{"title":"PMAR: A Lagrangian approach to the modelling of anthropogenic pressures for marine management","authors":"Sofia Bosi , Alessandra Raffaetà , Marta Simeoni , Nikola Bobchev , Dimitar Berov , Andrea Barbanti , Stefano Menegon","doi":"10.1016/j.envsoft.2025.106822","DOIUrl":"10.1016/j.envsoft.2025.106822","url":null,"abstract":"<div><div>PMAR (Pressure models for MARine activities) is a modelling framework and open-source Python-based software designed to assess anthropogenic pressures for marine management. PMAR uses Lagrangian trajectories calculated from ocean models to simulate pressure dispersion. An explicit link between pressures and their sources is established through a weight-based mechanism, which allows to rapidly explore different pressure source scenarios. Here, PMAR is adopted to investigate the distribution of surface macroplastics in the Black Sea under two scenarios, yielding results consistent with previous modelling studies and observations. Over 50% of the macroplastics in each Black Sea country after a yearly run came from elsewhere, stressing the importance of cross-boundary cooperation. Compared to other pressure modelling frameworks, PMAR emerges as a balanced compromise between computational efficiency and predictive accuracy. Future work will focus on making PMAR accessible to decision-makers and aligning it with requirements of Digital Twin of the Ocean applications.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106822"},"PeriodicalIF":4.6,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145689376","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 : 2025-12-04DOI: 10.1016/j.envsoft.2025.106825
Tommaso Baggio , Maximiliano Costa , Niccolò Marchi , Tommaso Locatelli , Emanuele Lingua
Windstorms are the primary cause of damage to European forests. Although different mechanistic and probabilistic models have been developed to estimate the vulnerability of forests to wind, their practical application remains limited. This study presents a new, semi-automated methodology for deriving tree and forest characteristics over large areas through the analysis of Canopy Height Model (CHM) data. By integrating the semi-mechanistic model ForestGALES, the developed algorithm uses these data to calculate spatially explicit maps of Critical Wind Speed (CWS). The presented methodology is applied to a real case study to calculate the CWS of forests in the Italian Eastern Alps. Results show that adding detailed and spatially distributed forest cover information improves the CWS calculations, thereby enhancing the reliability of models to assess forest wind vulnerability. Forest practitioners can take advantage of this new methodology to enhance the resistance and resilience of their forests through specific management techniques.
{"title":"Improve the estimation of forest wind vulnerability through remote sensed data: a new methodology","authors":"Tommaso Baggio , Maximiliano Costa , Niccolò Marchi , Tommaso Locatelli , Emanuele Lingua","doi":"10.1016/j.envsoft.2025.106825","DOIUrl":"10.1016/j.envsoft.2025.106825","url":null,"abstract":"<div><div>Windstorms are the primary cause of damage to European forests. Although different mechanistic and probabilistic models have been developed to estimate the vulnerability of forests to wind, their practical application remains limited. This study presents a new, semi-automated methodology for deriving tree and forest characteristics over large areas through the analysis of Canopy Height Model (CHM) data. By integrating the semi-mechanistic model ForestGALES, the developed algorithm uses these data to calculate spatially explicit maps of Critical Wind Speed (CWS). The presented methodology is applied to a real case study to calculate the CWS of forests in the Italian Eastern Alps. Results show that adding detailed and spatially distributed forest cover information improves the CWS calculations, thereby enhancing the reliability of models to assess forest wind vulnerability. Forest practitioners can take advantage of this new methodology to enhance the resistance and resilience of their forests through specific management techniques.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106825"},"PeriodicalIF":4.6,"publicationDate":"2025-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145689375","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 : 2025-12-04DOI: 10.1016/j.envsoft.2025.106819
Haiyang Shi , Ximing Cai , Xinchen Hu , Alaa Jamal , Donghui Li , Chao Sun , Xin-Zhong Liang
Irrigation accounts for a significant share of global freshwater use, and optimizing scheduling is crucial for improving water use efficiency. Current methods rely on short-term weather forecasts, limiting long-term planning. Additionally, most models are site-specific due to data constraints, and lacking national applicability. This study develops a real-time irrigation scheduling tool for cornfields across the Contiguous United States (CONUS). By integrating sub-seasonal to seasonal (S2S) climate forecasts with the Soil Water Atmosphere Plant (SWAP) model, the tool optimizes irrigation scheduling at any day in the season, balancing water cost and crop yield. A human-computer interaction framework provides real-time irrigation recommendations while incorporating farmer feedback. S2S-informed scheduling improves water use efficiency and net profit compared to default SWAP schedules. Various up-to-date CONUS-scale datasets helped to reduce dependence on in-situ observations and extend the applicability of the tool to diverse field conditions in the CONUS.
{"title":"A sub-seasonal to seasonal climate forecast informed irrigation scheduling tool for the Contiguous United States","authors":"Haiyang Shi , Ximing Cai , Xinchen Hu , Alaa Jamal , Donghui Li , Chao Sun , Xin-Zhong Liang","doi":"10.1016/j.envsoft.2025.106819","DOIUrl":"10.1016/j.envsoft.2025.106819","url":null,"abstract":"<div><div>Irrigation accounts for a significant share of global freshwater use, and optimizing scheduling is crucial for improving water use efficiency. Current methods rely on short-term weather forecasts, limiting long-term planning. Additionally, most models are site-specific due to data constraints, and lacking national applicability. This study develops a real-time irrigation scheduling tool for cornfields across the Contiguous United States (CONUS). By integrating sub-seasonal to seasonal (S2S) climate forecasts with the Soil Water Atmosphere Plant (SWAP) model, the tool optimizes irrigation scheduling at any day in the season, balancing water cost and crop yield. A human-computer interaction framework provides real-time irrigation recommendations while incorporating farmer feedback. S2S-informed scheduling improves water use efficiency and net profit compared to default SWAP schedules. Various up-to-date CONUS-scale datasets helped to reduce dependence on in-situ observations and extend the applicability of the tool to diverse field conditions in the CONUS.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106819"},"PeriodicalIF":4.6,"publicationDate":"2025-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145689378","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 : 2025-12-02DOI: 10.1016/j.envsoft.2025.106803
Zaharaddeeen Karami Lawal , Hayati Yassin , Daphne Teck Ching Lai , Azam Che Idris
Accurate significant wave height (SWH) forecasting is pivotal for maritime safety, coastal engineering, and offshore operations. This study explores three advanced deep learning models: Ensemble Patch-TST, LSTM with Attention Mechanism (SWH-AT-LSTM), and GRU with Attention Mechanism (SWH-AT-GRU) for SWH forecasting across multiple forecast horizons. The innovative Ensemble Patch-TST model combines the strengths of ensemble learning and transfer learning to achieve state-of-the-art performance, particularly in medium- and long-term predictions. Using datasets from four distinct offshore stations, the study evaluates each model's ability to capture temporal dependencies and generalize across diverse environmental conditions. Results demonstrate that the Ensemble Patch-TST outperforms baseline Patch-TST and other state-of-the-art models, achieving superior accuracy and robustness, particularly for datasets with limited training samples. This work not only highlights the transformative potential of advanced deep learning techniques in enhancing maritime forecasting systems but also emphasizes their scalability and efficiency, providing a practical and effective approach for real-world applications.
准确的有效波高（SWH）预报对海上安全、海岸工程和海上作业至关重要。本研究探索了三种先进的深度学习模型：集成Patch-TST、带注意机制的LSTM （SWH- at -LSTM）和带注意机制的GRU (SWH- at -GRU)，用于多预测视界的SWH预测。创新的集成补丁- tst模型结合了集成学习和迁移学习的优势，以实现最先进的性能，特别是在中长期预测方面。利用来自四个不同海上站点的数据集，该研究评估了每个模型捕获时间依赖性和在不同环境条件下进行推广的能力。结果表明，集成Patch-TST优于基线Patch-TST和其他最先进的模型，实现了卓越的准确性和鲁棒性，特别是对于有限训练样本的数据集。这项工作不仅突出了先进的深度学习技术在增强海事预报系统方面的变革潜力，而且强调了它们的可扩展性和效率，为现实世界的应用提供了一种实用而有效的方法。
{"title":"Optimizing significant wave height forecasting through Ensemble Patch-TST and attention-enhanced recurrent models","authors":"Zaharaddeeen Karami Lawal , Hayati Yassin , Daphne Teck Ching Lai , Azam Che Idris","doi":"10.1016/j.envsoft.2025.106803","DOIUrl":"10.1016/j.envsoft.2025.106803","url":null,"abstract":"<div><div>Accurate significant wave height (SWH) forecasting is pivotal for maritime safety, coastal engineering, and offshore operations. This study explores three advanced deep learning models: Ensemble Patch-TST, LSTM with Attention Mechanism (SWH-AT-LSTM), and GRU with Attention Mechanism (SWH-AT-GRU) for SWH forecasting across multiple forecast horizons. The innovative Ensemble Patch-TST model combines the strengths of ensemble learning and transfer learning to achieve state-of-the-art performance, particularly in medium- and long-term predictions. Using datasets from four distinct offshore stations, the study evaluates each model's ability to capture temporal dependencies and generalize across diverse environmental conditions. Results demonstrate that the Ensemble Patch-TST outperforms baseline Patch-TST and other state-of-the-art models, achieving superior accuracy and robustness, particularly for datasets with limited training samples. This work not only highlights the transformative potential of advanced deep learning techniques in enhancing maritime forecasting systems but also emphasizes their scalability and efficiency, providing a practical and effective approach for real-world applications.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106803"},"PeriodicalIF":4.6,"publicationDate":"2025-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145657347","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}
This study advances an integrated Bayesian support vector machine-based two-step factorial analysis (abbreviated as BSVM-TFA) method for revealing the influences of human activities on water demand. The developed method can capture complex nonlinear relationships between human activities and water demand by calibrating SVM hyperparameters through Bayesian optimization, which helps prevent overfitting. BSVM-TFA can also identify the individual and interactive effects of multiple factors on water demand and screen key influencing factors. The BSVM-TFA is then applied to Central Asia, and the results show that by 2050, water demand would range from 75.66 × 109 m3 to 113.23 × 109 m3 under different scenarios, indicating an uncertainty of about 33.18 % driven by human activities. The key factors influencing water demand in Central Asia are GDP and agricultural irrigation efficiency (AIE), with a total contribution of 47.98 %; the water demand would be reduced by 16.42 × 109 m3 with low-growth GDP and increasing AIE.
{"title":"Bayesian-factorial analysis for unveiling multi-factor interactive effect on water demand in Central Asia","authors":"Yanxiao Zhou , Yongping Li , Guohe Huang , Zhenyao Shen , Yufei Zhang","doi":"10.1016/j.envsoft.2025.106806","DOIUrl":"10.1016/j.envsoft.2025.106806","url":null,"abstract":"<div><div>This study advances an integrated Bayesian support vector machine-based two-step factorial analysis (abbreviated as BSVM-TFA) method for revealing the influences of human activities on water demand. The developed method can capture complex nonlinear relationships between human activities and water demand by calibrating SVM hyperparameters through Bayesian optimization, which helps prevent overfitting. BSVM-TFA can also identify the individual and interactive effects of multiple factors on water demand and screen key influencing factors. The BSVM-TFA is then applied to Central Asia, and the results show that by 2050, water demand would range from 75.66 × 10<sup>9</sup> m<sup>3</sup> to 113.23 × 10<sup>9</sup> m<sup>3</sup> under different scenarios, indicating an uncertainty of about 33.18 % driven by human activities. The key factors influencing water demand in Central Asia are GDP and agricultural irrigation efficiency (AIE), with a total contribution of 47.98 %; the water demand would be reduced by 16.42 × 10<sup>9</sup> m<sup>3</sup> with low-growth GDP and increasing AIE.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106806"},"PeriodicalIF":4.6,"publicationDate":"2025-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145619718","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 : 2025-11-29DOI: 10.1016/j.envsoft.2025.106805
Huanyu Yang , Hongming Zhang , Yuwei Sun , Lu Du , Weilin Xu , Jincheng Ni , Qiankun Chen , Chunmei Wang , Qinke Yang , Haijing Shi
Accurately extracting watershed boundaries is critical for hydrological modeling and environmental management. Traditional extraction methods from Digital Elevation Models (DEMs) rely on manually defined thresholds and supplementary terrain features, limiting adaptability and efficiency. To address these issues, this study developed a watershed boundaries extraction framework based on a Residual Bottleneck Attention Multi-feature Fusion Network (RBM-SegNet). The framework consists of three components: an input layer, a semantic segmentation model, and a post-processing module. Key contributions include: (1) utilizing the [DEM, Slope, Hillshade, and Aspect] functions as the optimal input combination; (2) introducing residual connections and the Bottleneck Attention Module (BAM) to enhance feature transmission and suppress irrelevant regions; (3) incorporating multi-feature fusion to refine structural and detail prediction; and (4) incorporating post-processing to improve output-completeness and hydrological consistency. The experimental results show that RBM-SegNet outperforms traditional and existing deep learning methods in accuracy, demonstrating strong potential for practical applications.
准确提取流域边界对水文建模和环境管理至关重要。传统的数字高程模型（dem）提取方法依赖于人工定义的阈值和补充地形特征，限制了适应性和效率。为了解决这些问题，本研究开发了一种基于残余瓶颈注意力多特征融合网络（RBM-SegNet）的分水岭边界提取框架。该框架由三个部分组成：输入层、语义分割模型和后处理模块。主要贡献包括：(1)利用[DEM， Slope, Hillshade， and Aspect]函数作为最优输入组合；(2)引入残差连接和瓶颈注意模块（BAM），增强特征传输，抑制不相关区域；(3)结合多特征融合对结构和细节预测进行精细化；(4)结合后处理，提高输出的完整性和水文一致性。实验结果表明，RBM-SegNet在准确性上优于传统和现有的深度学习方法，具有很强的实际应用潜力。
{"title":"Watershed boundary extraction from digital elevation models using RBM-SegNet","authors":"Huanyu Yang , Hongming Zhang , Yuwei Sun , Lu Du , Weilin Xu , Jincheng Ni , Qiankun Chen , Chunmei Wang , Qinke Yang , Haijing Shi","doi":"10.1016/j.envsoft.2025.106805","DOIUrl":"10.1016/j.envsoft.2025.106805","url":null,"abstract":"<div><div>Accurately extracting watershed boundaries is critical for hydrological modeling and environmental management. Traditional extraction methods from Digital Elevation Models (DEMs) rely on manually defined thresholds and supplementary terrain features, limiting adaptability and efficiency. To address these issues, this study developed a watershed boundaries extraction framework based on a Residual Bottleneck Attention Multi-feature Fusion Network (RBM-SegNet). The framework consists of three components: an input layer, a semantic segmentation model, and a post-processing module. Key contributions include: (1) utilizing the [DEM, Slope, Hillshade, and Aspect] functions as the optimal input combination; (2) introducing residual connections and the Bottleneck Attention Module (BAM) to enhance feature transmission and suppress irrelevant regions; (3) incorporating multi-feature fusion to refine structural and detail prediction; and (4) incorporating post-processing to improve output-completeness and hydrological consistency. The experimental results show that RBM-SegNet outperforms traditional and existing deep learning methods in accuracy, demonstrating strong potential for practical applications.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106805"},"PeriodicalIF":4.6,"publicationDate":"2025-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145613575","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 : 2025-11-28DOI: 10.1016/j.envsoft.2025.106802
Xin Jing, Xue Yang, JunGang Luo, GangGang Zuo
Integrating deep learning with hydrological models is a popular research direction; however, this field faces significant challenges due to automatic differentiation requirements and interface incompatibilities, leading to many existing hydrological modeling frameworks being unable to perform effective hybrid modeling. To fill this gap, we propose a framework that inherits and enhances the design philosophies of previous modeling frameworks. It utilizes symbolic programming to reduce the difficulty of hydrological modeling, particularly for hybrid models integrating deep learning, supports automatic differentiation for model optimization, and effectively addresses the diverse and evolving needs for both specialized hydrological and hybrid modeling applications. This framework, named HydroModels.jl, is implemented in the Julia programming language, is publicly accessible on GitHub, and is accompanied by detailed documentation. This study describes its architecture and implementation details, and presents two case studies as examples to demonstrate its integration capabilities and applicability.
{"title":"A flexible, differentiable framework for neural-enhanced hydrological modeling: Design, implementation, and applications with HydroModels.jl","authors":"Xin Jing, Xue Yang, JunGang Luo, GangGang Zuo","doi":"10.1016/j.envsoft.2025.106802","DOIUrl":"10.1016/j.envsoft.2025.106802","url":null,"abstract":"<div><div>Integrating deep learning with hydrological models is a popular research direction; however, this field faces significant challenges due to automatic differentiation requirements and interface incompatibilities, leading to many existing hydrological modeling frameworks being unable to perform effective hybrid modeling. To fill this gap, we propose a framework that inherits and enhances the design philosophies of previous modeling frameworks. It utilizes symbolic programming to reduce the difficulty of hydrological modeling, particularly for hybrid models integrating deep learning, supports automatic differentiation for model optimization, and effectively addresses the diverse and evolving needs for both specialized hydrological and hybrid modeling applications. This framework, named HydroModels.jl, is implemented in the Julia programming language, is publicly accessible on GitHub, and is accompanied by detailed documentation. This study describes its architecture and implementation details, and presents two case studies as examples to demonstrate its integration capabilities and applicability.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"197 ","pages":"Article 106802"},"PeriodicalIF":4.6,"publicationDate":"2025-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145613576","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 : 2025-11-27DOI: 10.1016/j.envsoft.2025.106799
Jamiu Adekunle Idowu , Ayman Alfahid
Floods are among the world's most devastating hazards, yet progress in predicting and managing flood risk remains limited by pervasive uncertainties at every stage of the modelling pipeline. This systematic review identifies eight open problems in uncertainty quantification for flood modelling, including: long-term prediction errors, poor calibration of predictive intervals, incomplete representation of uncertainties, inadequate handling of spatial and temporal variability, non-linearity, data scarcity and integration issues, high computational costs, and failure to capture uncertainties in extreme events. These challenges reflect a system-level mismatch between the dynamic complexity of floods and the fragmented nature of current modelling practice. Real progress in flood risk science demands a shift from siloed, modular workflows to seamless, end-to-end probabilistic pipelines – integrating heterogeneous data, hybridizing process-based and data-driven models, rigorously quantifying uncertainty at all stages, and communicating actionable risk information for policy and emergency response.
{"title":"Open problems in uncertainty quantification for flood modelling: A systematic review","authors":"Jamiu Adekunle Idowu , Ayman Alfahid","doi":"10.1016/j.envsoft.2025.106799","DOIUrl":"10.1016/j.envsoft.2025.106799","url":null,"abstract":"<div><div>Floods are among the world's most devastating hazards, yet progress in predicting and managing flood risk remains limited by pervasive uncertainties at every stage of the modelling pipeline. This systematic review identifies eight open problems in uncertainty quantification for flood modelling, including: long-term prediction errors, poor calibration of predictive intervals, incomplete representation of uncertainties, inadequate handling of spatial and temporal variability, non-linearity, data scarcity and integration issues, high computational costs, and failure to capture uncertainties in extreme events. These challenges reflect a system-level mismatch between the dynamic complexity of floods and the fragmented nature of current modelling practice. Real progress in flood risk science demands a shift from siloed, modular workflows to seamless, end-to-end probabilistic pipelines – integrating heterogeneous data, hybridizing process-based and data-driven models, rigorously quantifying uncertainty at all stages, and communicating actionable risk information for policy and emergency response.</div></div>","PeriodicalId":310,"journal":{"name":"Environmental Modelling & Software","volume":"196 ","pages":"Article 106799"},"PeriodicalIF":4.6,"publicationDate":"2025-11-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145611928","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