Ehsan Roshani, Pavel Popov, Yehuda Kleiner, Sina Sanjari, Andrew Colombo, Mostafa Bigdeli
Intentional chemical contamination of water distribution systems (WDSs) could have severe health consequences. High potency chemicals constituting, in essence, ‘super poisons’ have the potential to be used in such intrusion scenarios. Some of these contaminants are capable of killing the victim in less than hour. Due to their high toxicity levels and short period of time from exposure to the onset of symptoms, 9-1-1 call centers are likely the first point of contact for the victims or their families with the authorities. Information such as 9-1-1 calls could be used to identify the ongoing event and potential intrusion locations. In this way, such emergency calls could function as an intrusion warning system. This study employs network hydraulic modeling to synthesize the 9-1-1 call patterns in the aftermath of such events. It then defines the scenarios as a multi-label pattern recognition problem. The synthesized data then was used to train a convolutional neural network (CNN). The trained artificial intelligence (AI), was applied to a real-world WDS with approximately 4,000 km of pipe and 26,000 demand nodes. The results indicated that the CNN is capable of accurately recognizing the pattern and pinpointing the originating location of the intrusion with an accuracy greater than 93%.
{"title":"Detecting and locating chemical intrusion in water distribution systems using 9-1-1 calls","authors":"Ehsan Roshani, Pavel Popov, Yehuda Kleiner, Sina Sanjari, Andrew Colombo, Mostafa Bigdeli","doi":"10.2166/hydro.2024.299","DOIUrl":"https://doi.org/10.2166/hydro.2024.299","url":null,"abstract":"\u0000 \u0000 Intentional chemical contamination of water distribution systems (WDSs) could have severe health consequences. High potency chemicals constituting, in essence, ‘super poisons’ have the potential to be used in such intrusion scenarios. Some of these contaminants are capable of killing the victim in less than hour. Due to their high toxicity levels and short period of time from exposure to the onset of symptoms, 9-1-1 call centers are likely the first point of contact for the victims or their families with the authorities. Information such as 9-1-1 calls could be used to identify the ongoing event and potential intrusion locations. In this way, such emergency calls could function as an intrusion warning system. This study employs network hydraulic modeling to synthesize the 9-1-1 call patterns in the aftermath of such events. It then defines the scenarios as a multi-label pattern recognition problem. The synthesized data then was used to train a convolutional neural network (CNN). The trained artificial intelligence (AI), was applied to a real-world WDS with approximately 4,000 km of pipe and 26,000 demand nodes. The results indicated that the CNN is capable of accurately recognizing the pattern and pinpointing the originating location of the intrusion with an accuracy greater than 93%.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140376116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
According to the European Directive 2007/60/CE, flood risk evaluation should include a cost–benefit analysis (CBA) on a long-term time horizon to evaluate the impact of mitigation measures. The standard CBA assumes to know in advance the events observed in the time horizon and a priori compares all mitigation measures by an economic metric. No change is supposed to be made to these measures throughout the time horizon. This modus operandi is not appropriate in the domain of flood risk management because several conditions are uncertain when the CBA is made (e.g., urban policies). This article faces these challenges by the integration of cost–benefit analysis and decision trees, to prescribe mitigation measures under uncertainty on the budget for mitigation actions because their funding can be modified after the conclusion of the CBA. The former integration is discussed in the real case of the lowland valley of the Coghinas River (Sardinia, Italy), for which the classical CBA compared five mitigation measures of infrastructural works. The integration into the decision tree also allows to evaluating mitigation measures with changes in infrastructural works and a lamination action. The outcomes advise to decreasing the maximum storage level and increase the peak lamination.
{"title":"Decision trees in cost–benefit analysis for flood risk management plans","authors":"J. Napolitano, Massimo Di Francesco, G. Sechi","doi":"10.2166/hydro.2024.194","DOIUrl":"https://doi.org/10.2166/hydro.2024.194","url":null,"abstract":"\u0000 According to the European Directive 2007/60/CE, flood risk evaluation should include a cost–benefit analysis (CBA) on a long-term time horizon to evaluate the impact of mitigation measures. The standard CBA assumes to know in advance the events observed in the time horizon and a priori compares all mitigation measures by an economic metric. No change is supposed to be made to these measures throughout the time horizon. This modus operandi is not appropriate in the domain of flood risk management because several conditions are uncertain when the CBA is made (e.g., urban policies). This article faces these challenges by the integration of cost–benefit analysis and decision trees, to prescribe mitigation measures under uncertainty on the budget for mitigation actions because their funding can be modified after the conclusion of the CBA. The former integration is discussed in the real case of the lowland valley of the Coghinas River (Sardinia, Italy), for which the classical CBA compared five mitigation measures of infrastructural works. The integration into the decision tree also allows to evaluating mitigation measures with changes in infrastructural works and a lamination action. The outcomes advise to decreasing the maximum storage level and increase the peak lamination.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140382794","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Due to global climate change, managing water resources is one of the most critical challenges for most countries in the world, especially in the Middle East. In the Kurdistan Region of Iraq (KRI), there is a good amount of precipitation, surface-water, and groundwater, but the main issue is mismanagement of those sources. Rainfall is one of the major sources of water resources in KRI. In order to manage the available water resources and prevent natural disasters such as floods and droughts, there is a need for reliable models for forecasting rainfall. The current study focuses on developing a hybrid model namely, seasonal autoregressive integrated moving average combined with an artificial neural network (SARIMA-ANN) for forecasting monthly rainfall at Sulaymaniyah City for the duration of 1938–2012. For comparison purposes, a conventional machine learning model, namely, artificial neural networks (ANN) has been applied on the same data. Two different statistical measurements, namely, root mean square error (RMSE) and coefficient of determination (R2) have been used to check the accuracy of the proposed models. According to the findings, SARIMA-ANN outperformed ANN with RMSE = 11.5, RMSE = 51.002, R2 = 0.98, R2 = 0.43, respectively. The findings of the current study could contribute to sustainable development goal (SDG) 6.
{"title":"Developing an innovative machine learning model for rainfall prediction in a Semi-Arid region","authors":"S. Latif, Dyar Othman Mohammed, Alhassan Jaafar","doi":"10.2166/hydro.2024.014","DOIUrl":"https://doi.org/10.2166/hydro.2024.014","url":null,"abstract":"\u0000 Due to global climate change, managing water resources is one of the most critical challenges for most countries in the world, especially in the Middle East. In the Kurdistan Region of Iraq (KRI), there is a good amount of precipitation, surface-water, and groundwater, but the main issue is mismanagement of those sources. Rainfall is one of the major sources of water resources in KRI. In order to manage the available water resources and prevent natural disasters such as floods and droughts, there is a need for reliable models for forecasting rainfall. The current study focuses on developing a hybrid model namely, seasonal autoregressive integrated moving average combined with an artificial neural network (SARIMA-ANN) for forecasting monthly rainfall at Sulaymaniyah City for the duration of 1938–2012. For comparison purposes, a conventional machine learning model, namely, artificial neural networks (ANN) has been applied on the same data. Two different statistical measurements, namely, root mean square error (RMSE) and coefficient of determination (R2) have been used to check the accuracy of the proposed models. According to the findings, SARIMA-ANN outperformed ANN with RMSE = 11.5, RMSE = 51.002, R2 = 0.98, R2 = 0.43, respectively. The findings of the current study could contribute to sustainable development goal (SDG) 6.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140223054","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Supriya Chaudhary, G. Singh, Deepak Gupta, Suruchi Singh Maunas, V. Mishra
This study includes groundwater quality data from 290 monitoring sites from 69 districts of Uttar Pradesh, India. The analysis of the data showed that 1.3, 75.52, 47.93, and 31.03% of groundwater samples had concentrations of electrical conductivity (EC), total hardness (TH), Mg2+, and HCO3−, respectively, higher than the maximum permissible limit. Groundwater quality index (GWQI) was calculated for these 290 monitoring sites which revealed that 21 sites (7.24%) had inappropriate GWQI for drinking water, and 18 sites (7.24%) had an unsuitable index for irrigation. Most of the sampling sites (98.97%) showed high EC contents in groundwater with a mean value of 999.33 μS/cm. Fluoride content was found within the permissible limits in 95.52% of the samples, while 4.48% had high concentrations. The use of hierarchical cluster analysis differentiated all the sites into two clusters: one with high pollution and the other with low pollution. Significant correlations exist between physicochemical and irrigation indicators in the correlation matrix. High loadings of EC, TH, Ca2+, Mg2+, Na+, Cl−, and SO42− were identified in the first principal component, which are thought to be pollution-controlled processes from anthropogenic sources. According to the Chadha diagram, CaHCO3 and Ca–Mg–HCl were the two most prevalent chemicals in the water.
{"title":"Characterization of groundwater potability and irrigation potential in Uttar Pradesh, India using water quality index and multivariate statistics","authors":"Supriya Chaudhary, G. Singh, Deepak Gupta, Suruchi Singh Maunas, V. Mishra","doi":"10.2166/hydro.2024.291","DOIUrl":"https://doi.org/10.2166/hydro.2024.291","url":null,"abstract":"\u0000 \u0000 This study includes groundwater quality data from 290 monitoring sites from 69 districts of Uttar Pradesh, India. The analysis of the data showed that 1.3, 75.52, 47.93, and 31.03% of groundwater samples had concentrations of electrical conductivity (EC), total hardness (TH), Mg2+, and HCO3−, respectively, higher than the maximum permissible limit. Groundwater quality index (GWQI) was calculated for these 290 monitoring sites which revealed that 21 sites (7.24%) had inappropriate GWQI for drinking water, and 18 sites (7.24%) had an unsuitable index for irrigation. Most of the sampling sites (98.97%) showed high EC contents in groundwater with a mean value of 999.33 μS/cm. Fluoride content was found within the permissible limits in 95.52% of the samples, while 4.48% had high concentrations. The use of hierarchical cluster analysis differentiated all the sites into two clusters: one with high pollution and the other with low pollution. Significant correlations exist between physicochemical and irrigation indicators in the correlation matrix. High loadings of EC, TH, Ca2+, Mg2+, Na+, Cl−, and SO42− were identified in the first principal component, which are thought to be pollution-controlled processes from anthropogenic sources. According to the Chadha diagram, CaHCO3 and Ca–Mg–HCl were the two most prevalent chemicals in the water.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140231084","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Water diversion projects in high-latitude areas often reduce the risk of ice jams in winter by reducing the water transfer flow, which might cause the waste of water transfer benefits. This paper establishes a real-time prediction system of water temperature in winter, which can predict the change in water temperature by inputting the air temperature forecast data and the current hydraulic data. Taking the middle route of the south-to-north water diversion project as the background, the model parameters calibration and system application testing at different time periods are carried out. The results show that the prediction errors of water temperature for the 1- and 7-day are relatively small, and the prediction errors of water temperature at four observation stations can be controlled within ±0.3 and ±0.6 °C, with the root mean square error (RMSE) ranging from 0.07 to 0.25 and 0.12 to 0.36, respectively. The 15-day water temperature prediction results are greatly affected by air temperature input conditions. The prediction errors for the first 7 days are relatively small, ranging from −0.59 to 0.36 °C, and the errors for the last 8 days increase as the accuracy of the air temperature forecast decreases, ranging from −2.42 to 0.22 °C.
{"title":"Application of real-time water temperature prediction system in winter for long-distance water diversion projects","authors":"Zepeng Xu, Mengkai Liu, Minghai Huang, Letian Wen, Xinlei Guo","doi":"10.2166/hydro.2024.064","DOIUrl":"https://doi.org/10.2166/hydro.2024.064","url":null,"abstract":"\u0000 \u0000 Water diversion projects in high-latitude areas often reduce the risk of ice jams in winter by reducing the water transfer flow, which might cause the waste of water transfer benefits. This paper establishes a real-time prediction system of water temperature in winter, which can predict the change in water temperature by inputting the air temperature forecast data and the current hydraulic data. Taking the middle route of the south-to-north water diversion project as the background, the model parameters calibration and system application testing at different time periods are carried out. The results show that the prediction errors of water temperature for the 1- and 7-day are relatively small, and the prediction errors of water temperature at four observation stations can be controlled within ±0.3 and ±0.6 °C, with the root mean square error (RMSE) ranging from 0.07 to 0.25 and 0.12 to 0.36, respectively. The 15-day water temperature prediction results are greatly affected by air temperature input conditions. The prediction errors for the first 7 days are relatively small, ranging from −0.59 to 0.36 °C, and the errors for the last 8 days increase as the accuracy of the air temperature forecast decreases, ranging from −2.42 to 0.22 °C.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140234727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pavlos Vryoni Pavlou, S. Filippou, Solon Solonos, Stelios G. Vrachimis, Kleanthis Malialis, Demetrios G. Eliades, Theocharis Theocarides, Marios M. Polycarpou
Monitoring the water usage of different appliances and informing consumers about it has been shown to have an impact on their behavior toward drinking water conservation. The most practical and cost-effective way to accomplish this is through a non-intrusive approach, that locally analyzes data received from a flow sensor at the main water supply pipe of a household. In this work, we present two different methods addressing the challenges of disaggregating end-use consumption and classifying consumption events. The first method is model-based (MB) and uses a combination of dynamic time wrapping and statistical bounds to analyze four water end-use characteristics. The second, learning-based (LB) method is data-driven and formulates the problem as a time-series classification problem without relying on a priori identification of events. We perform an extensive computational study that includes a comparison between an MB and an LB method, as well as an experimental study to demonstrate the application of the LB method on an edge computing device. Both methods achieve similar F1 scores (LB: 71.73%, MB: 71.04%) with the LB being more precise. The embedded LB method achieves a slightly higher score (72.01%) while enhancing on-site real-time processing, improving security, and privacy and enabling cost savings.
{"title":"Monitoring domestic water consumption: a comparative study of model-based and data-driven end-use disaggregation methods","authors":"Pavlos Vryoni Pavlou, S. Filippou, Solon Solonos, Stelios G. Vrachimis, Kleanthis Malialis, Demetrios G. Eliades, Theocharis Theocarides, Marios M. Polycarpou","doi":"10.2166/hydro.2024.120","DOIUrl":"https://doi.org/10.2166/hydro.2024.120","url":null,"abstract":"\u0000 Monitoring the water usage of different appliances and informing consumers about it has been shown to have an impact on their behavior toward drinking water conservation. The most practical and cost-effective way to accomplish this is through a non-intrusive approach, that locally analyzes data received from a flow sensor at the main water supply pipe of a household. In this work, we present two different methods addressing the challenges of disaggregating end-use consumption and classifying consumption events. The first method is model-based (MB) and uses a combination of dynamic time wrapping and statistical bounds to analyze four water end-use characteristics. The second, learning-based (LB) method is data-driven and formulates the problem as a time-series classification problem without relying on a priori identification of events. We perform an extensive computational study that includes a comparison between an MB and an LB method, as well as an experimental study to demonstrate the application of the LB method on an edge computing device. Both methods achieve similar F1 scores (LB: 71.73%, MB: 71.04%) with the LB being more precise. The embedded LB method achieves a slightly higher score (72.01%) while enhancing on-site real-time processing, improving security, and privacy and enabling cost savings.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140233349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Physics-based numerical models often depend on several parameters to close. Some of them can be expressed using established theoretical or empirical closure formulations. However, some others aggregate complex physical processes and are hence left as tuneable parameters, and can only be calibrated by trial and error. Yet, calibration data are not always available to do so, which prevents these models from being applied to wide ranges of laboratory or river flows. We hence propose a machine learning-based methodology to close any group of unclosed and correlated parameters, applied here to a two-phase/two-layer (2P2L) morphodynamical model. The methodology combines a numerical experiment with a known theoretical result and machine learning. It is applied to the considered model to close two friction parameters for which generalizable and vastly acknowledged closure formulations lack in the literature. The resulting hybrid model, combining the original 2P2L model and the closure models, is tested against two laboratory dam break test cases. Despite excessive smoothness and underestimation of the concentration in sediment, the hybrid model performed similarly to other models from the literature requiring trial and error calibration and showed high stability and accuracy regarding the estimation of the water-sediment mixture's inertia.
{"title":"Machine learning rather than trial and error to close morphodynamical tuneable parameters: application to a two-phase/two-layer model","authors":"R. Meurice, S. Soares-Frazão","doi":"10.2166/hydro.2024.183","DOIUrl":"https://doi.org/10.2166/hydro.2024.183","url":null,"abstract":"\u0000 \u0000 Physics-based numerical models often depend on several parameters to close. Some of them can be expressed using established theoretical or empirical closure formulations. However, some others aggregate complex physical processes and are hence left as tuneable parameters, and can only be calibrated by trial and error. Yet, calibration data are not always available to do so, which prevents these models from being applied to wide ranges of laboratory or river flows. We hence propose a machine learning-based methodology to close any group of unclosed and correlated parameters, applied here to a two-phase/two-layer (2P2L) morphodynamical model. The methodology combines a numerical experiment with a known theoretical result and machine learning. It is applied to the considered model to close two friction parameters for which generalizable and vastly acknowledged closure formulations lack in the literature. The resulting hybrid model, combining the original 2P2L model and the closure models, is tested against two laboratory dam break test cases. Despite excessive smoothness and underestimation of the concentration in sediment, the hybrid model performed similarly to other models from the literature requiring trial and error calibration and showed high stability and accuracy regarding the estimation of the water-sediment mixture's inertia.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140243070","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The significance of improving rainfall prediction methods has escalated due to climate change-induced flash floods and severe flooding. In this study, rainfall nowcasting has been studied utilizing NASA Giovanni (Goddard Interactive Online Visualization and Analysis Infrastructure) satellite-derived precipitation products and the convolutional long short-term memory (ConvLSTM) approach. The goal of the study is to assess the impact of data augmentation on flood nowcasting. Due to data requirements of deep learning-based prediction methods, data augmentation is performed using eight different interpolation techniques. Spatial, temporal, and spatio-temporal interpolated rainfall data are used to conduct a comparative analysis of the results obtained through nowcasting rainfall. This research examines two catastrophic floods that transpired in the Türkiye Marmara Region in 2009 and the Central Black Sea Region in 2021, which are selected as the focal case studies. The Marmara and Black Sea regions are prone to frequent flooding, which, due to the dense population, has devastating consequences. Furthermore, these regions exhibit distinct topographical characteristics and precipitation patterns, and the frontal systems that impact them are also dissimilar. The nowcast results for the two regions exhibit a significant difference. Although data augmentation significantly reduced the error values by 59% for one region, it did not yield the same effectiveness for the other region.
由于气候变化导致的山洪暴发和严重洪灾,改进降雨预测方法的意义日益凸显。本研究利用 NASA Giovanni(戈达德交互式在线可视化和分析基础设施)卫星降水产品和卷积长短期记忆(ConvLSTM)方法对降雨预报进行了研究。研究的目的是评估数据增强对洪水预报的影响。由于基于深度学习的预测方法对数据有要求,因此使用八种不同的插值技术进行数据扩增。使用空间、时间和时空插值降雨数据,对通过降雨预报获得的结果进行比较分析。本研究选择 2009 年在土耳其马尔马拉地区和 2021 年在黑海中部地区发生的两次灾难性洪灾作为重点案例研究。马尔马拉和黑海地区经常发生洪灾,由于人口密集,洪灾造成了破坏性后果。此外,这两个地区的地形特点和降水模式截然不同,影响它们的锋面系统也不尽相同。这两个地区的预报结果显示出显著差异。虽然数据扩增将一个地区的误差值大幅降低了 59%,但对另一个地区却没有产生同样的效果。
{"title":"Temporal and spatial satellite data augmentation for deep learning-based rainfall nowcasting","authors":"Özlem Baydaroğlu Yeşilköy, I. Demir","doi":"10.2166/hydro.2024.235","DOIUrl":"https://doi.org/10.2166/hydro.2024.235","url":null,"abstract":"\u0000 The significance of improving rainfall prediction methods has escalated due to climate change-induced flash floods and severe flooding. In this study, rainfall nowcasting has been studied utilizing NASA Giovanni (Goddard Interactive Online Visualization and Analysis Infrastructure) satellite-derived precipitation products and the convolutional long short-term memory (ConvLSTM) approach. The goal of the study is to assess the impact of data augmentation on flood nowcasting. Due to data requirements of deep learning-based prediction methods, data augmentation is performed using eight different interpolation techniques. Spatial, temporal, and spatio-temporal interpolated rainfall data are used to conduct a comparative analysis of the results obtained through nowcasting rainfall. This research examines two catastrophic floods that transpired in the Türkiye Marmara Region in 2009 and the Central Black Sea Region in 2021, which are selected as the focal case studies. The Marmara and Black Sea regions are prone to frequent flooding, which, due to the dense population, has devastating consequences. Furthermore, these regions exhibit distinct topographical characteristics and precipitation patterns, and the frontal systems that impact them are also dissimilar. The nowcast results for the two regions exhibit a significant difference. Although data augmentation significantly reduced the error values by 59% for one region, it did not yield the same effectiveness for the other region.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140247051","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Estuaries are ecologically valuable regions where tidal forces move large volumes of water. To understand the ongoing physical processes in such dynamic systems, a series of estuarine monitoring stations is required. Based on the measurements, estuarine dynamics can be described by key values, so-called tidal characteristics. The reconstruction and prediction of tidal characteristics by suitable approaches is essential to discover natural or anthropogenic changes. Therefore, it is of interest to inter- and extrapolate measured values in time and to investigate the spatial relationship between different stations. Normally, such system analyses are performed by deterministic numerical models. However, to facilitate long-term investigations also, statistical and machine learning approaches are good options. For a Weser estuary case study, we implemented three approaches (linear, non-linear, and artificial neural network regression) with the same database to enable the prediction of tidal extrema. Thereby we achieve an accuracy of 0.4–2.5% derivation (based on the RMSEs) while approximating measured values over 19 years. This proves that the approaches can be used for hindcast studies as well as for future analysis of system changes. Our work can be understood as a proof of concept for the practical potential of neural networks in estuarine system analysis.
{"title":"Using statistical and machine learning approaches to describe estuarine tidal dynamics","authors":"Franziska Lauer, Frank Kösters","doi":"10.2166/hydro.2024.294","DOIUrl":"https://doi.org/10.2166/hydro.2024.294","url":null,"abstract":"\u0000 Estuaries are ecologically valuable regions where tidal forces move large volumes of water. To understand the ongoing physical processes in such dynamic systems, a series of estuarine monitoring stations is required. Based on the measurements, estuarine dynamics can be described by key values, so-called tidal characteristics. The reconstruction and prediction of tidal characteristics by suitable approaches is essential to discover natural or anthropogenic changes. Therefore, it is of interest to inter- and extrapolate measured values in time and to investigate the spatial relationship between different stations. Normally, such system analyses are performed by deterministic numerical models. However, to facilitate long-term investigations also, statistical and machine learning approaches are good options. For a Weser estuary case study, we implemented three approaches (linear, non-linear, and artificial neural network regression) with the same database to enable the prediction of tidal extrema. Thereby we achieve an accuracy of 0.4–2.5% derivation (based on the RMSEs) while approximating measured values over 19 years. This proves that the approaches can be used for hindcast studies as well as for future analysis of system changes. Our work can be understood as a proof of concept for the practical potential of neural networks in estuarine system analysis.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140245654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Christiaan Schutte, M. van der Laan, B. J. van der Merwe
Streamflow information is crucial for effectively managing water resources. The declining number of active gauging stations in many rivers is a global concern, necessitating the need for reliable streamflow estimates. Deep learning techniques offer potential solutions, but their application in southern Africa remains largely underexplored. To fill this gap, this study evaluated the predictive performance of gated recurrent unit (GRU) and long short-term memory (LSTM) networks using two headwater catchments of the Steelpoort River, South Africa, as case studies. The model inputs included rainfall, maximum, and minimum temperature, as well as past streamflow, which was utilized in an autoregressive sense. The inclusion of streamflow in this way allowed for the incorporation of simulated streamflow values into the look-back window for predicting the streamflow of the testing set. Two modifications were required to the GRU and LSTM architectures to ensure physically consistent predictions, including a change in the activation function of the GRU/LSTM cells in the final hidden layer, and a non-negative constraint that was used in the dense layer. Models trained using commercial weather station data produced reliable streamflow estimates, while moderately accurate predictions were obtained using freely available gridded weather data.
{"title":"Leveraging historic streamflow and weather data with deep learning for enhanced streamflow predictions","authors":"Christiaan Schutte, M. van der Laan, B. J. van der Merwe","doi":"10.2166/hydro.2024.268","DOIUrl":"https://doi.org/10.2166/hydro.2024.268","url":null,"abstract":"\u0000 \u0000 Streamflow information is crucial for effectively managing water resources. The declining number of active gauging stations in many rivers is a global concern, necessitating the need for reliable streamflow estimates. Deep learning techniques offer potential solutions, but their application in southern Africa remains largely underexplored. To fill this gap, this study evaluated the predictive performance of gated recurrent unit (GRU) and long short-term memory (LSTM) networks using two headwater catchments of the Steelpoort River, South Africa, as case studies. The model inputs included rainfall, maximum, and minimum temperature, as well as past streamflow, which was utilized in an autoregressive sense. The inclusion of streamflow in this way allowed for the incorporation of simulated streamflow values into the look-back window for predicting the streamflow of the testing set. Two modifications were required to the GRU and LSTM architectures to ensure physically consistent predictions, including a change in the activation function of the GRU/LSTM cells in the final hidden layer, and a non-negative constraint that was used in the dense layer. Models trained using commercial weather station data produced reliable streamflow estimates, while moderately accurate predictions were obtained using freely available gridded weather data.","PeriodicalId":507813,"journal":{"name":"Journal of Hydroinformatics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140417210","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}