Pub Date : 2026-06-01Epub Date: 2026-01-17DOI: 10.1016/j.pce.2026.104308
Ayoub Ghazzar, Abdelkader Larabi, Mohamed Jalal El Hamidi
Groundwater is one of the most valuable natural resources, particularly for coastal regions that depend on it for freshwater supply. However, under the influence of Climate Change (CC), groundwater is becoming increasingly scarce and vulnerable to contamination, either through anthropogenic activities or natural processes, such as seawater intrusion (SWI). This article reviews several studies that investigate the dynamics of SWI using lab-scale physical models. Researchers have adopted various experimental approaches, including modifications to the dimensions of physical models, selection of different materials for model components, and simulation of natural phenomena such as Sea Level Rise (SLR), tidal fluctuations, and variations in porous media homogeneity and heterogeneity. Additionally, human activities have been simulated through freshwater pumping experiments to assess their impact on saltwater intrusion dynamics. The studies reviewed in this article have employed both visual observations and electrical resistivity tomography (ERT) to monitor SWI processes, often complementing experimental work with numerical simulations. By analyzing and comparing these different methodologies, this review article provides insights into the strengths and limitations of various approaches, offering a comprehensive perspective on laboratory-scale investigations of SWI in coastal aquifers and serving as a practical guide for future research.
{"title":"A review of lab-scale physical models for SWI: From sharp to density-driven interface, with an analysis of model components and influencing physical factors","authors":"Ayoub Ghazzar, Abdelkader Larabi, Mohamed Jalal El Hamidi","doi":"10.1016/j.pce.2026.104308","DOIUrl":"10.1016/j.pce.2026.104308","url":null,"abstract":"<div><div>Groundwater is one of the most valuable natural resources, particularly for coastal regions that depend on it for freshwater supply. However, under the influence of Climate Change (CC), groundwater is becoming increasingly scarce and vulnerable to contamination, either through anthropogenic activities or natural processes, such as seawater intrusion (SWI). This article reviews several studies that investigate the dynamics of SWI using lab-scale physical models. Researchers have adopted various experimental approaches, including modifications to the dimensions of physical models, selection of different materials for model components, and simulation of natural phenomena such as Sea Level Rise (SLR), tidal fluctuations, and variations in porous media homogeneity and heterogeneity. Additionally, human activities have been simulated through freshwater pumping experiments to assess their impact on saltwater intrusion dynamics. The studies reviewed in this article have employed both visual observations and electrical resistivity tomography (ERT) to monitor SWI processes, often complementing experimental work with numerical simulations. By analyzing and comparing these different methodologies, this review article provides insights into the strengths and limitations of various approaches, offering a comprehensive perspective on laboratory-scale investigations of SWI in coastal aquifers and serving as a practical guide for future research.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104308"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146015538","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-01-13DOI: 10.1016/j.pce.2026.104304
Kanak N. Moharir , Chaitanya Baliram Pande , Abebe Debele Tolche , Abhay M. Varade , Saad Shauket Sammen , Mohammad Khalid , Krishna Kumar Yadav , Maha Awjan Alreshidi , Ghadah Shukri Albakri , Mohamed Elsahabi
In this paper, we have focused on the sustainable water and soil conservation important for the development of ecosystem planning. The assessment of soil erosion is important study for the basaltic rock area in India. Therefore, to delineation of groundwater potential zones (GPZ) mapping was done for the sustainable soil erosion conservation, and integrated planning with the help of the aquifer mapping, remote sensing (RS), analytical hierarchical process (AHP), and geographic information system (GIS). The data integration process methods are useful to identify soil erosion risk and delineating GPZ and soil erosion mapping. In this study, total seven layers such as slope, land use/land cover (LULC), soil, geology, geomorphology, drainage density and lineament density was used for suitable analysis of the GPZ mapping. These seven thematic layers were assigned weights using AHP and GIS methods in Arc GIS 10.5 software with multiple-criteria decision analysis (MCDA) techniques. The integration of seven layers provides valuable insights into managing and sustaining groundwater resources. We have identified the five classes such as very low, low, medium, high, and very high in GPZ map. The results are found the conservation area is 83 % and other area under 17 % in the study area. The aquifer mapping results help to understanding the groundwater resources in the basaltic rock. The current research outcomes are develop the groundwater potential zones map and soil conservation plans based on thematic layers, GIS system and methods. This paper results helpful to future mitigating of water risks, preventing soil erosion, addressing water scarcity, managing climate risks, and improving drought conditions. These results of study area will support to planning and management of groundwater and natural resources, which can helpful to local government administrators, researchers, and planners for making policy in the improvement of groundwater resources.
{"title":"Development of groundwater potential zones and soil erosion mapping with planning based on aquifer modeling, AHP and geospatial techniques","authors":"Kanak N. Moharir , Chaitanya Baliram Pande , Abebe Debele Tolche , Abhay M. Varade , Saad Shauket Sammen , Mohammad Khalid , Krishna Kumar Yadav , Maha Awjan Alreshidi , Ghadah Shukri Albakri , Mohamed Elsahabi","doi":"10.1016/j.pce.2026.104304","DOIUrl":"10.1016/j.pce.2026.104304","url":null,"abstract":"<div><div>In this paper, we have focused on the sustainable water and soil conservation important for the development of ecosystem planning. The assessment of soil erosion is important study for the basaltic rock area in India. Therefore, to delineation of groundwater potential zones (GPZ) mapping was done for the sustainable soil erosion conservation, and integrated planning with the help of the aquifer mapping, remote sensing (RS), analytical hierarchical process (AHP), and geographic information system (GIS). The data integration process methods are useful to identify soil erosion risk and delineating GPZ and soil erosion mapping. In this study, total seven layers such as slope, land use/land cover (LULC), soil, geology, geomorphology, drainage density and lineament density was used for suitable analysis of the GPZ mapping. These seven thematic layers were assigned weights using AHP and GIS methods in Arc GIS 10.5 software with multiple-criteria decision analysis (MCDA) techniques. The integration of seven layers provides valuable insights into managing and sustaining groundwater resources. We have identified the five classes such as very low, low, medium, high, and very high in GPZ map. The results are found the conservation area is 83 % and other area under 17 % in the study area. The aquifer mapping results help to understanding the groundwater resources in the basaltic rock. The current research outcomes are develop the groundwater potential zones map and soil conservation plans based on thematic layers, GIS system and methods. This paper results helpful to future mitigating of water risks, preventing soil erosion, addressing water scarcity, managing climate risks, and improving drought conditions. These results of study area will support to planning and management of groundwater and natural resources, which can helpful to local government administrators, researchers, and planners for making policy in the improvement of groundwater resources.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104304"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146081258","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-01-27DOI: 10.1016/j.pce.2026.104319
Tarekegn Dejen Mengistu , Il-Moon Chung , Sun Woo Chang
Effective water quality monitoring requires predictive models that combine high accuracy, interpretability, and credible uncertainty quantification. Machine learning (ML) techniques have emerged as powerful tools for predicting water quality and quantifying associated uncertainties. Similarly, Bayesian deep learning (BDL) frameworks enable probabilistic predictions that quantify uncertainties. These approaches can capture nonlinear interactions and provide robust predictions in diverse environmental conditions. This study integrated ensemble ML and BDL to assess the complex relationships between physicochemical parameters and the Water Quality Index (WQI). Six supervised ensemble ML algorithms, namely Decision Tree (DT), Random Forest (RF), Extra Trees (ERT), XGBoost, CatBoost, and LightGBM were evaluated using Bayesian optimisation to identify the optimal hyperparameter configurations. DT achieved the highest predictive accuracy with MAE = 0.657 and 0.428, RMSE = 1.181 and 0.747, MAPE = 10.561 and 7.155, R2 = 0.960 and 0.987, and nRMSE = 0.065 and 0.042 for the training and test sets, respectively. The DT outperformed more complex ensemble models, and SHapley Additive exPlanations (SHAP)-based eXplainable Artificial Intelligence (XAI) identified the most influential predictors, aligning model predictions with underlying hydrochemical processes. To capture predictive uncertainty, a probabilistic BDL was developed yielding probabilistic outputs and explicit epistemic uncertainty estimates. ROC analysis confirmed strong performance across WQI classes, with AUC scores of up to 0.90 for WQI classes. The probabilistic approach provides actionable insights for adaptive water quality management, enabling targeted monitoring in areas of high uncertainty and supporting transparent, evidence-based decision-making. These results underscore the value of integrating ML, and Bayesian optimisation to advance robust and adaptive water quality assessment. The proposed workflow provides a scalable framework to enhance monitoring, optimize resources, and advance sustainable water management aligned with the SDGs.
{"title":"Machine learning for water quality prediction and uncertainty assessment","authors":"Tarekegn Dejen Mengistu , Il-Moon Chung , Sun Woo Chang","doi":"10.1016/j.pce.2026.104319","DOIUrl":"10.1016/j.pce.2026.104319","url":null,"abstract":"<div><div>Effective water quality monitoring requires predictive models that combine high accuracy, interpretability, and credible uncertainty quantification. Machine learning (ML) techniques have emerged as powerful tools for predicting water quality and quantifying associated uncertainties. Similarly, Bayesian deep learning (BDL) frameworks enable probabilistic predictions that quantify uncertainties. These approaches can capture nonlinear interactions and provide robust predictions in diverse environmental conditions. This study integrated ensemble ML and BDL to assess the complex relationships between physicochemical parameters and the Water Quality Index (WQI). Six supervised ensemble ML algorithms, namely Decision Tree (DT), Random Forest (RF), Extra Trees (ERT), XGBoost, CatBoost, and LightGBM were evaluated using Bayesian optimisation to identify the optimal hyperparameter configurations. DT achieved the highest predictive accuracy with MAE = 0.657 and 0.428, RMSE = 1.181 and 0.747, MAPE = 10.561 and 7.155, R<sup>2</sup> = 0.960 and 0.987, and nRMSE = 0.065 and 0.042 for the training and test sets, respectively. The DT outperformed more complex ensemble models, and SHapley Additive exPlanations (SHAP)-based eXplainable Artificial Intelligence (XAI) identified the most influential predictors, aligning model predictions with underlying hydrochemical processes. To capture predictive uncertainty, a probabilistic BDL was developed yielding probabilistic outputs and explicit epistemic uncertainty estimates. ROC analysis confirmed strong performance across WQI classes, with AUC scores of up to 0.90 for WQI classes. The probabilistic approach provides actionable insights for adaptive water quality management, enabling targeted monitoring in areas of high uncertainty and supporting transparent, evidence-based decision-making. These results underscore the value of integrating ML, and Bayesian optimisation to advance robust and adaptive water quality assessment. The proposed workflow provides a scalable framework to enhance monitoring, optimize resources, and advance sustainable water management aligned with the SDGs.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104319"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-01-24DOI: 10.1016/j.pce.2026.104317
Bijay Halder , Biswarup Rana , Liew Juneng , Malay Pramanik , Chaitanya Baliram Pande , Samyah Salem Refadah , Mohd Yawar Ali Khan , Shafik S. Shafik , Zaher Mundher Yaseen
Climate change affects urbanisation in several distinct ways. Beyond the evident temperature fluctuations, it triggers urbanisation impacts by causing natural disasters, changing energy structure, and the geographical distribution of precipitation. The effect of surface urban heat islands (SUHI) has become increasingly noticeable in Thailand. Therefore, this study used Landsat datasets and the Google Earth Engine (GEE) cloud-based platform to examine land use change, geospatial indices, heat-island effects, and temperature fluctuations from 2015 to 2023. The LULC change (LULCC) recorded vegetation decreased by −40.40 km2, agricultural land by −48.73 km2, water bodies by −6.90 km2, and the built-up land increased by 75.28 km2 from 2015 to 2023. The Bangkok Metropolitan Administration (BMA) area and its adjacent regions are identified most urbanized regions. The land surface temperature (LST) rose from 39.24 °C (2015) to 40.12 °C (2023), and similarly, SUHI increased by 0.166 to 1.223. Major cities (e.g., Bangkok, Pattaya, Phuket, Chiang Mai, and Hat Yai) recorded the most significant reductions in the vegetation index (0.03) and development in the built-up index (0.04). The analysis is valuable for working adaptation strategies that measure risk associated with present climate change effects, which are impacted by hydrometeorological variables such as precipitation, temperature, humidity, and evaporation.
{"title":"Urbanization-climate change Interactions and their effects on surface heat island in tropical monsoon environments","authors":"Bijay Halder , Biswarup Rana , Liew Juneng , Malay Pramanik , Chaitanya Baliram Pande , Samyah Salem Refadah , Mohd Yawar Ali Khan , Shafik S. Shafik , Zaher Mundher Yaseen","doi":"10.1016/j.pce.2026.104317","DOIUrl":"10.1016/j.pce.2026.104317","url":null,"abstract":"<div><div>Climate change affects urbanisation in several distinct ways. Beyond the evident temperature fluctuations, it triggers urbanisation impacts by causing natural disasters, changing energy structure, and the geographical distribution of precipitation. The effect of surface urban heat islands (SUHI) has become increasingly noticeable in Thailand. Therefore, this study used Landsat datasets and the Google Earth Engine (GEE) cloud-based platform to examine land use change, geospatial indices, heat-island effects, and temperature fluctuations from 2015 to 2023. The LULC change (LULCC) recorded vegetation decreased by −40.40 km<sup>2</sup>, agricultural land by −48.73 km<sup>2</sup>, water bodies by −6.90 km<sup>2</sup>, and the built-up land increased by 75.28 km<sup>2</sup> from 2015 to 2023. The Bangkok Metropolitan Administration (BMA) area and its adjacent regions are identified most urbanized regions. The land surface temperature (LST) rose from 39.24 °C (2015) to 40.12 °C (2023), and similarly, SUHI increased by 0.166 to 1.223. Major cities (e.g., Bangkok, Pattaya, Phuket, Chiang Mai, and Hat Yai) recorded the most significant reductions in the vegetation index (0.03) and development in the built-up index (0.04). The analysis is valuable for working adaptation strategies that measure risk associated with present climate change effects, which are impacted by hydrometeorological variables such as precipitation, temperature, humidity, and evaporation.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104317"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174698","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Timely and accurate flood detection is critical for minimizing loss of life and assessing disaster-induced damage. Remote sensing technologies enable consistent, near–real-time mapping of flood extents, which is essential for effective disaster response and preparedness. Among these, Synthetic Aperture Radar (SAR) data have proven particularly valuable due to their all-weather, day–night imaging capability. This study presents a robust and computationally efficient deep learning framework for flood mapping using Sentinel-1 SAR imagery. Using the Sen1Floods11 benchmark dataset, which encompasses 11 geographically and temporally diverse flood events, we propose DeeplabV3PlusMX (DB-SEN1FloodNet)—a ‘Maxout’ enhanced semantic segmentation model derived from the DeeplabV3+ architecture. The proposed model strategically integrates ‘Maxout’ activation layers at multiple stages of the encoder–decoder pipeline to reduce feature-space redundancy, enhance robustness to speckle noise, and improve generalization across heterogeneous flood scenarios. By enabling adaptive piecewise linear feature representations and synergistic regularization with dropout, the ‘Maxout’ strategy improves discrimination of flooded areas under complex scattering conditions. Experimental results demonstrate strong performance, achieving an overall accuracy of 96%, average precision of 98%, mean recall of 94%, an F1-score of 96%, and a mean Intersection over Union (mIoU) of 65%. Furthermore, the model exhibits superior transferability when evaluated on unseen geographic regions, outperforming existing approaches that are often limited to event- or region-specific training. The reduced reliance on auxiliary datasets further underscores the operational potential of the proposed framework for scalable, global flood monitoring using SAR data.
{"title":"A Maxout-enhanced robust deep convolutional neural network model for flood mapping using Sentinel-1 SAR data","authors":"Shubham Awasthi , Gopal Singh Phartiyal , Divyesh Varade , Kamal Jain","doi":"10.1016/j.pce.2026.104316","DOIUrl":"10.1016/j.pce.2026.104316","url":null,"abstract":"<div><div>Timely and accurate flood detection is critical for minimizing loss of life and assessing disaster-induced damage. Remote sensing technologies enable consistent, near–real-time mapping of flood extents, which is essential for effective disaster response and preparedness. Among these, Synthetic Aperture Radar (SAR) data have proven particularly valuable due to their all-weather, day–night imaging capability. This study presents a robust and computationally efficient deep learning framework for flood mapping using Sentinel-1 SAR imagery. Using the Sen1Floods11 benchmark dataset, which encompasses 11 geographically and temporally diverse flood events, we propose DeeplabV3PlusMX (DB-SEN1FloodNet)—a ‘Maxout’ enhanced semantic segmentation model derived from the DeeplabV3+ architecture. The proposed model strategically integrates ‘Maxout’ activation layers at multiple stages of the encoder–decoder pipeline to reduce feature-space redundancy, enhance robustness to speckle noise, and improve generalization across heterogeneous flood scenarios. By enabling adaptive piecewise linear feature representations and synergistic regularization with dropout, the ‘Maxout’ strategy improves discrimination of flooded areas under complex scattering conditions. Experimental results demonstrate strong performance, achieving an overall accuracy of 96%, average precision of 98%, mean recall of 94%, an F1-score of 96%, and a mean Intersection over Union (mIoU) of 65%. Furthermore, the model exhibits superior transferability when evaluated on unseen geographic regions, outperforming existing approaches that are often limited to event- or region-specific training. The reduced reliance on auxiliary datasets further underscores the operational potential of the proposed framework for scalable, global flood monitoring using SAR data.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104316"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174281","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
This paper provides a systematic review that synthesizes recent advancements in the integration of machine learning (ML) and remote sensing techniques for flood modelling in developing regions between 2010 and 2025. To achieve the main objective the study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) to search for articles in key databases such as Google Scholar, Web of Science and Science Direct. A total of 50197 articles were obtained and screened based on specific set criterion and a total of 126 articles were obtained after screening and were used in this study. These articles were then subjected to bibliometric analysis which revealed an exponential growth in the number of articles obtained with a sharp rise in publications post-2018. Further analysis revealed that most studies concentrated in South and East Asia, highlighting regional bias and underrepresentation of data-scarce areas such as Africa and Latin America. The results further indicated a widespread increase in the use of freely available remote sensing data (e.g., Sentinel-1/2, Landsat, MODIS) which was driven by accessibility and resolution advantages, while advanced but cost-prohibitive platforms (e.g., RADARSAT, UAVs) remain underutilized. Comparative analysis of model performance showed that traditional hydrological and hydraulic models remain relevant but often suffer from oversimplification and high data and computational demands. In contrast, ML models like CNNs, RF, and SVM demonstrated robust performance with AUC values frequently exceeding 0.90. However, the lack of consistent benchmarking, standardized evaluation metrics, and open-source codebases limits model comparability and reproducibility across studies. Furthermore, most reviewed studies overlook uncertainty quantification, compound event interactions, and tail dependence. To address these gaps, the review recommends integrating uncertainty-aware techniques such as Bayesian deep learning (e.g., MC-Dropout) and copula-based bivariate extreme value models. Moreover, emphasis should be placed on the ethical deployment of ML in flood-prone regions, advocating for transparency in model assumptions, fairness assessments, and participatory model design. Future research should prioritize scalable, interpretable, and equitable modelling approaches, particularly in underrepresented and high-risk regions.
本文系统综述了2010年至2025年发展中地区洪水建模中机器学习（ML）和遥感技术集成的最新进展。为了达到主要目的，本研究遵循系统评价和元分析首选报告项目（PRISMA）在谷歌Scholar、Web of Science和Science Direct等关键数据库中检索文章。根据特定的设定标准，共获得50197篇文章进行筛选，筛选后共获得126篇文章用于本研究。然后对这些文章进行文献计量学分析，结果显示，2018年后，随着出版物的急剧增加，获得的文章数量呈指数级增长。进一步的分析表明，大多数研究集中在南亚和东亚，突出了区域偏见和非洲和拉丁美洲等数据稀缺地区的代表性不足。结果进一步表明，由于可及性和分辨率优势，免费获得的遥感数据（如Sentinel-1/2、Landsat、MODIS）的使用广泛增加，而先进但成本过高的平台（如RADARSAT、无人机）仍未得到充分利用。模型性能的对比分析表明，传统的水文和水力模型仍然具有相关性，但往往存在过度简化和数据量和计算量大的问题。相比之下，像cnn、RF和SVM这样的ML模型表现出鲁棒性，AUC值经常超过0.90。然而，缺乏一致的基准测试、标准化的评估指标和开源代码库限制了模型的可比性和跨研究的可重复性。此外，大多数综述的研究忽略了不确定性量化、复合事件相互作用和尾部依赖性。为了解决这些差距，该综述建议整合不确定性感知技术，如贝叶斯深度学习（例如MC-Dropout）和基于copula的二元极值模型。此外，重点应放在洪水易发地区ML的道德部署上，倡导模型假设、公平评估和参与式模型设计的透明度。未来的研究应优先考虑可扩展、可解释和公平的建模方法，特别是在代表性不足和高风险地区。
{"title":"Integrating remote sensing and machine learning for flood modelling: A systematic literature review","authors":"Naledi Manyaka , Cletah Shoko , Siyamthanda Gxokwe , Timothy Dube","doi":"10.1016/j.pce.2026.104315","DOIUrl":"10.1016/j.pce.2026.104315","url":null,"abstract":"<div><div>This paper provides a systematic review that synthesizes recent advancements in the integration of machine learning (ML) and remote sensing techniques for flood modelling in developing regions between 2010 and 2025. To achieve the main objective the study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) to search for articles in key databases such as Google Scholar, Web of Science and Science Direct. A total of 50197 articles were obtained and screened based on specific set criterion and a total of 126 articles were obtained after screening and were used in this study. These articles were then subjected to bibliometric analysis which revealed an exponential growth in the number of articles obtained with a sharp rise in publications post-2018. Further analysis revealed that most studies concentrated in South and East Asia, highlighting regional bias and underrepresentation of data-scarce areas such as Africa and Latin America. The results further indicated a widespread increase in the use of freely available remote sensing data (e.g., Sentinel-1/2, Landsat, MODIS) which was driven by accessibility and resolution advantages, while advanced but cost-prohibitive platforms (e.g., RADARSAT, UAVs) remain underutilized. Comparative analysis of model performance showed that traditional hydrological and hydraulic models remain relevant but often suffer from oversimplification and high data and computational demands. In contrast, ML models like CNNs, RF, and SVM demonstrated robust performance with AUC values frequently exceeding 0.90. However, the lack of consistent benchmarking, standardized evaluation metrics, and open-source codebases limits model comparability and reproducibility across studies. Furthermore, most reviewed studies overlook uncertainty quantification, compound event interactions, and tail dependence. To address these gaps, the review recommends integrating uncertainty-aware techniques such as Bayesian deep learning (e.g., MC-Dropout) and copula-based bivariate extreme value models. Moreover, emphasis should be placed on the ethical deployment of ML in flood-prone regions, advocating for transparency in model assumptions, fairness assessments, and participatory model design. Future research should prioritize scalable, interpretable, and equitable modelling approaches, particularly in underrepresented and high-risk regions.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104315"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-01-09DOI: 10.1016/j.pce.2026.104277
Saeedreza Moazeni , Ali Salajegheh , Shahram Khalighi-Sigaroodi , Ali Golkarian , Artemi Cerda
Watersheds serve as fundamental units of hydrological systems, playing a crucial role in freshwater supply, biodiversity conservation, and ecosystem sustainability. This study conducts a bibliometric and qualitative analysis of global research (1979–2024) to evaluate the effectiveness of watershed management interventions on key water-balance components, including runoff, infiltration, evapotranspiration, and groundwater recharge. Findings indicate a significant increase in publications, particularly after 2011. We used VOSviewer and Bibliometrix to map publication trends, co-occurrence and co-citation networks, leading authors and journals, and thematic clusters. Prominent studies focus on Land-use impacts on water resources, the application of hydrological models, and the role of vegetation in water regulation. The United States, China, and Canada lead research efforts in this field. Our synthesis classifies interventions into biological, mechanical and managerial types and summarizes reported effects on runoff, infiltration, evapotranspiration (ET), soil moisture, surface storage and groundwater recharge. Results show a marked increase in publications after 2011 and concentration of research activity in the United States, China and Canada. Among water-balance components, runoff (n = 447) and ET (n = 308) dominate the literature, whereas soil moisture (n = 66), surface storage (n = 40) and baseflow (n = 38) are comparatively underrepresented. Hydrological process models were the most used estimation approach (n = 440), followed by remote sensing (n = 140), groundwater models (n = 82) and machine-learning methods (n = 38). From the reviewed studies we synthesise typical outcomes: mechanical structures (e.g., terraces, check-dams) are frequently associated with reductions in surface runoff (up to ∼25 % in reported cases) and context-dependent increases in recharge (reported ranges of ∼40–70 %), while biological measures (e.g., afforestation) often improve infiltration but can elevate ET in water-limited environments. We identify recurrent methodological shortcomings — inconsistent reporting of uncertainty, limited reproducibility of bibliometric settings, and scarce comparative field studies — and propose a focused research agenda: transparent bibliometric reporting, prioritized monitoring of underexplored components (soil moisture, baseflow), development of hybrid process–data modeling frameworks, and targeted, context-specific evaluations of interventions under climate variability. This synthesis provides a state-of-the-art overview and a structured set of priorities to guide future watershed management research and policy.
{"title":"Effectiveness of watershed management on water balance components-a review","authors":"Saeedreza Moazeni , Ali Salajegheh , Shahram Khalighi-Sigaroodi , Ali Golkarian , Artemi Cerda","doi":"10.1016/j.pce.2026.104277","DOIUrl":"10.1016/j.pce.2026.104277","url":null,"abstract":"<div><div>Watersheds serve as fundamental units of hydrological systems, playing a crucial role in freshwater supply, biodiversity conservation, and ecosystem sustainability. This study conducts a bibliometric and qualitative analysis of global research (1979–2024) to evaluate the effectiveness of watershed management interventions on key water-balance components, including runoff, infiltration, evapotranspiration, and groundwater recharge. Findings indicate a significant increase in publications, particularly after 2011. We used VOSviewer and Bibliometrix to map publication trends, co-occurrence and co-citation networks, leading authors and journals, and thematic clusters. Prominent studies focus on Land-use impacts on water resources, the application of hydrological models, and the role of vegetation in water regulation. The United States, China, and Canada lead research efforts in this field. Our synthesis classifies interventions into biological, mechanical and managerial types and summarizes reported effects on runoff, infiltration, evapotranspiration (ET), soil moisture, surface storage and groundwater recharge. Results show a marked increase in publications after 2011 and concentration of research activity in the United States, China and Canada. Among water-balance components, runoff (n = 447) and ET (n = 308) dominate the literature, whereas soil moisture (n = 66), surface storage (n = 40) and baseflow (n = 38) are comparatively underrepresented. Hydrological process models were the most used estimation approach (n = 440), followed by remote sensing (n = 140), groundwater models (n = 82) and machine-learning methods (n = 38). From the reviewed studies we synthesise typical outcomes: mechanical structures (e.g., terraces, check-dams) are frequently associated with reductions in surface runoff (up to ∼25 % in reported cases) and context-dependent increases in recharge (reported ranges of ∼40–70 %), while biological measures (e.g., afforestation) often improve infiltration but can elevate ET in water-limited environments. We identify recurrent methodological shortcomings — inconsistent reporting of uncertainty, limited reproducibility of bibliometric settings, and scarce comparative field studies — and propose a focused research agenda: transparent bibliometric reporting, prioritized monitoring of underexplored components (soil moisture, baseflow), development of hybrid process–data modeling frameworks, and targeted, context-specific evaluations of interventions under climate variability. This synthesis provides a state-of-the-art overview and a structured set of priorities to guide future watershed management research and policy.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104277"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174697","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-03DOI: 10.1016/j.pce.2026.104313
K.K. Rao , Mandira Barman , S.P. Datta , Debarup Das , V.K. Sharma , Debrup Ghosh , T.K. Das
The problem of phosphorus (P) fixation into insoluble mineral forms is particularly acute in the intensively cultivated Indo-Gangetic Plains (IGP), where decades of fertilizer application have created complex P dynamics. Although earlier studies have examined P availability in different soils, the mechanistic understanding of equilibrium relationships between soil-solution phosphate species and solid P phases under varying pH across the IGP remains limited.
To address this gap, the present study quantitatively assessed the chemical equilibria between soil-solution phosphate (H2PO4−) and dominant solid P phases by modelling soil solution and solid-phase interactions using geochemical equilibrium approaches across a representative pH gradient. This methodological framework enabled identification of the controlling mineral-phases and their saturation states under different soil reactions.
Results revealed that H2PO4− concentrations varied markedly with pH, ranging from 3901 × 10−6 to 397 × 10−6 M. Oversaturation of mineral phases such as variscite and K-taranakite was observed, particularly in soils receiving long-term applications of potassium- and ammonium-based phosphatic fertilizers. In acidic soils, P solubility was governed by iron- and aluminium-phosphate equilibria, whereas in neutral to alkaline soils, calcium phosphate phases predominated.
These findings demonstrate that soil pH and fertilizer legacy effects jointly regulate P solubility equilibria and phase transitions across the IGP. The study underscores the necessity of developing management strategies to mobilize the accumulated, less-available P pools through pH-specific interventions. Such approaches can enhance sustainable P use efficiency, reduce fertilizer dependency, and improve long-term crop productivity, contributing valuable insights to regional nutrient management and global P sustainability frameworks.
{"title":"Chemical equilibrium of solid phases governing phosphorus solubility in intensively cultivated soils of the Indo-Gangetic Plains","authors":"K.K. Rao , Mandira Barman , S.P. Datta , Debarup Das , V.K. Sharma , Debrup Ghosh , T.K. Das","doi":"10.1016/j.pce.2026.104313","DOIUrl":"10.1016/j.pce.2026.104313","url":null,"abstract":"<div><div>The problem of phosphorus (P) fixation into insoluble mineral forms is particularly acute in the intensively cultivated Indo-Gangetic Plains (IGP), where decades of fertilizer application have created complex P dynamics. Although earlier studies have examined P availability in different soils, the mechanistic understanding of equilibrium relationships between soil-solution phosphate species and solid P phases under varying pH across the IGP remains limited.</div><div>To address this gap, the present study quantitatively assessed the chemical equilibria between soil-solution phosphate (H<sub>2</sub>PO<sub>4</sub><sup>−</sup>) and dominant solid P phases by modelling soil solution and solid-phase interactions using geochemical equilibrium approaches across a representative pH gradient. This methodological framework enabled identification of the controlling mineral-phases and their saturation states under different soil reactions.</div><div>Results revealed that H<sub>2</sub>PO<sub>4</sub><sup>−</sup> concentrations varied markedly with pH, ranging from 3901 × 10<sup>−6</sup> to 397 × 10<sup>−6</sup> M. Oversaturation of mineral phases such as variscite and K-taranakite was observed, particularly in soils receiving long-term applications of potassium- and ammonium-based phosphatic fertilizers. In acidic soils, P solubility was governed by iron- and aluminium-phosphate equilibria, whereas in neutral to alkaline soils, calcium phosphate phases predominated.</div><div>These findings demonstrate that soil pH and fertilizer legacy effects jointly regulate P solubility equilibria and phase transitions across the IGP. The study underscores the necessity of developing management strategies to mobilize the accumulated, less-available P pools through pH-specific interventions. Such approaches can enhance sustainable P use efficiency, reduce fertilizer dependency, and improve long-term crop productivity, contributing valuable insights to regional nutrient management and global P sustainability frameworks.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104313"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174279","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-03DOI: 10.1016/j.pce.2026.104332
Andlia Abdoussalami , Dinesh Madhushanka , Xuesong Zhang , Qi Li , Zhenghua Hu , Abu Reza Md Towfiqul Islam
Tropical cyclones pose major risks to Small Island Developing States (SIDS) and their agriculture. Banana is one of the island's key subsistence and market crops, supporting household food security and local economies. As a vital crop, it holds significant importance both for daily consumption and economic exchange. This study provides the first island-level, crop-specific assessment of Cyclone Kenneth's impact on banana plantations across Ngazidja (Grande Comore). The Normalized Difference Vegetation Index (NDVI) is a spectral index derived from satellite imagery, which quantifies vegetation health. NDVI is calculated as the difference between near-infrared and red reflectance, providing a measure of canopy greenness. In this study, we used seasonally paired NDVI composites to reduce cloud contamination and account for short-term phenological variations between the pre- and post-cyclone imagery. Island-scale land-use analysis revealed a cropland decrease of −18.7% and a bare-land increase of +24.3%, while representative northern banana parcels (contiguous field of banana crops, delineated from high-resolution satellite imagery that is large enough to be represented accurately at a 10-m resolution) experienced severe canopy loss (ΔNDVI ≈ −0.32). Spatial regression identified proximity to the cyclone track, wind speed, and rainfall as the most significant predictors of vegetation loss, amplified in low-elevation and gentle-slope areas. The findings demonstrate the effectiveness of integrating multi-sensor remote sensing with exposure modeling for rapid post-cyclone agricultural damage assessment in data-limited island environments, supporting evidence-based recovery and resilience planning.
{"title":"Harnessing vegetation indices and remote sensing to assess the impact of Cyclone Kenneth on banana plantations: Insights from Ngazidja Island (Comoros)","authors":"Andlia Abdoussalami , Dinesh Madhushanka , Xuesong Zhang , Qi Li , Zhenghua Hu , Abu Reza Md Towfiqul Islam","doi":"10.1016/j.pce.2026.104332","DOIUrl":"10.1016/j.pce.2026.104332","url":null,"abstract":"<div><div>Tropical cyclones pose major risks to Small Island Developing States (SIDS) and their agriculture. Banana is one of the island's key subsistence and market crops, supporting household food security and local economies. As a vital crop, it holds significant importance both for daily consumption and economic exchange. This study provides the first island-level, crop-specific assessment of Cyclone Kenneth's impact on banana plantations across Ngazidja (Grande Comore). The Normalized Difference Vegetation Index (NDVI) is a spectral index derived from satellite imagery, which quantifies vegetation health. NDVI is calculated as the difference between near-infrared and red reflectance, providing a measure of canopy greenness. In this study, we used seasonally paired NDVI composites to reduce cloud contamination and account for short-term phenological variations between the pre- and post-cyclone imagery. Island-scale land-use analysis revealed a cropland decrease of −18.7% and a bare-land increase of +24.3%, while representative northern banana parcels (contiguous field of banana crops, delineated from high-resolution satellite imagery that is large enough to be represented accurately at a 10-m resolution) experienced severe canopy loss (ΔNDVI ≈ −0.32). Spatial regression identified proximity to the cyclone track, wind speed, and rainfall as the most significant predictors of vegetation loss, amplified in low-elevation and gentle-slope areas. The findings demonstrate the effectiveness of integrating multi-sensor remote sensing with exposure modeling for rapid post-cyclone agricultural damage assessment in data-limited island environments, supporting evidence-based recovery and resilience planning.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104332"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174275","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-06-01Epub Date: 2026-02-05DOI: 10.1016/j.pce.2026.104331
Mahdi Delghandi, Behnam Abdollah-Abadi
Probable Maximum Precipitation (PMP) is a fundamental input for estimating the Probable Maximum Flood (PMF), and therefore plays a key role in the safe and cost-effective design of hydraulic structures. However, the estimation of PMP is inherently associated with various sources of uncertainty. This study investigates uncertainties in PMP estimation in Iran, focusing on calculation methods (Hershfield method (H), the Site-Specific (SS) method, and a newly proposed Developed Site-Specific (DSS) approach), rain-gauge station density, and climatic conditions. Monte Carlo simulation and Sobol’ global sensitivity analysis were employed to quantify contribution of frequency factor curve parameters (α, c and LR) to the uncertainty in PMP estimates. Results revealed that H approach produced the highest PMP values (135-1175 mm), while DSS, most compatible with Iran's climate, yielded lower values (110-632 mm). Application of DSS reduced the weighted mean PMP from 309 mm to 230 mm, which can significantly reduce design costs of high-hazard hydraulic structures. Uncertainty analysis demonstrated that climatic conditions accounted for the largest share of total uncertainty (41.2%), followed by the number of stations (36.2%) and the PMP estimation method (22.6%), respectively. Regardless of the specific contribution of each source of uncertainty, it is evident that all three factors have a significant impact on PMP estimation. Sobol’ analysis demonstrated that parameter c is the dominant contributor to PMP uncertainty. These findings highlight the need for an uncertainty-aware framework to support cost-efficient and safe hydraulic infrastructure planning.
{"title":"Uncertainty quantification in the estimation of Probable Maximum Precipitation (PMP) in Iran: A comprehensive analysis","authors":"Mahdi Delghandi, Behnam Abdollah-Abadi","doi":"10.1016/j.pce.2026.104331","DOIUrl":"10.1016/j.pce.2026.104331","url":null,"abstract":"<div><div>Probable Maximum Precipitation (PMP) is a fundamental input for estimating the Probable Maximum Flood (PMF), and therefore plays a key role in the safe and cost-effective design of hydraulic structures. However, the estimation of PMP is inherently associated with various sources of uncertainty. This study investigates uncertainties in PMP estimation in Iran, focusing on calculation methods (Hershfield method (H), the Site-Specific (SS) method, and a newly proposed Developed Site-Specific (DSS) approach), rain-gauge station density, and climatic conditions. Monte Carlo simulation and Sobol’ global sensitivity analysis were employed to quantify contribution of frequency factor curve parameters (<em>α</em>, <em>c</em> and <em>LR</em>) to the uncertainty in PMP estimates. Results revealed that H approach produced the highest PMP values (135-1175 mm), while DSS, most compatible with Iran's climate, yielded lower values (110-632 mm). Application of DSS reduced the weighted mean PMP from 309 mm to 230 mm, which can significantly reduce design costs of high-hazard hydraulic structures. Uncertainty analysis demonstrated that climatic conditions accounted for the largest share of total uncertainty (41.2%), followed by the number of stations (36.2%) and the PMP estimation method (22.6%), respectively. Regardless of the specific contribution of each source of uncertainty, it is evident that all three factors have a significant impact on PMP estimation. Sobol’ analysis demonstrated that parameter <em>c</em> is the dominant contributor to PMP uncertainty. These findings highlight the need for an uncertainty-aware framework to support cost-efficient and safe hydraulic infrastructure planning.</div></div>","PeriodicalId":54616,"journal":{"name":"Physics and Chemistry of the Earth","volume":"143 ","pages":"Article 104331"},"PeriodicalIF":4.1,"publicationDate":"2026-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146174650","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"地球科学","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