Pub Date : 2023-12-07DOI: 10.5194/hess-27-4317-2023
R. Moua, N. Lesparre, Jean-François Girard, B. Belfort, F. Lehmann, Anis Younes
Abstract. In this study, we investigate the use of ground-penetrating radar (GPR) time-lapse monitoring of artificial soil infiltration experiments. The aim is to evaluate this protocol in the context of estimating the hydrodynamic unsaturated soil parameter values and their associated uncertainties. The originality of this work is to suggest a statistical parameter estimation approach using Markov chain Monte Carlo (MCMC) methods to have direct estimates of the parameter uncertainties. Using the GPR time data from the moving wetting front only does not provide reliable results. Thus, we propose to use additional information from other types of reflectors to optimize the quality of the parameter estimation. Water movement and electromagnetic wave propagation in the unsaturated zone are modeled using a one-dimensional hydrogeophysical model. The GPR travel time data are analyzed for different reflectors: a moving reflector (the infiltration wetting front) and three fixed reflectors located at different depths in the soil. Global sensitivity analysis (GSA) is employed to assess the influence of the saturated hydraulic conductivity Ks, the saturated and residual water contents θs and θr, and the Mualem–van Genuchten shape parameters α and n of the soil on the GPR travel time data of the reflectors. Statistical calibration of the soil parameters is then performed using the MCMC method. The impact of the type of reflector (moving or fixed) is then evaluated by analyzing the calibrated model parameters and their confidence intervals for different scenarios. GSA results show that the sensitivities of the GPR data to the hydrodynamic soil parameters are different between moving and fixed reflectors, whereas fixed reflectors at various depths have similar sensitivities. Ks has a similar and strong influence on the data of both types of reflectors. Concerning the other parameters, for the wetting front, only θs and α have an influence, and only at long time steps since the total variance is zero at the very beginning of the experiment. On the other hand, for the fixed reflectors, the total variance is not zero at the very start and the parameters θs, θr, α and n can have an influence from the very beginning of the infiltration. Results of parameter estimation show that the use of calibration data from the moving or fixed reflectors alone does not enable a good identification of all soil parameters. With the moving reflector, the error between the estimated mean value and the exact target value for θr and α is 9 % and 45 %, respectively, and less than 3 % for the other parameters. The best reduction of the size of the parameter distribution is obtained for n, with a posterior distribution 9 times smaller than the prior one. For the others, this reduction ratio varies between 1 and 5. For the fixed reflectors, the estimated mean values are far from the target values for α, θr and n, representing for a reflector located at 120 cm 15 %, 27 %, and 121 %, respec
{"title":"Coupled hydrogeophysical inversion of an artificial infiltration experiment monitored with ground-penetrating radar: synthetic demonstration","authors":"R. Moua, N. Lesparre, Jean-François Girard, B. Belfort, F. Lehmann, Anis Younes","doi":"10.5194/hess-27-4317-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4317-2023","url":null,"abstract":"Abstract. In this study, we investigate the use of ground-penetrating radar (GPR) time-lapse monitoring of artificial soil infiltration experiments. The aim is to evaluate this protocol in the context of estimating the hydrodynamic unsaturated soil parameter values and their associated uncertainties. The originality of this work is to suggest a statistical parameter estimation approach using Markov chain Monte Carlo (MCMC) methods to have direct estimates of the parameter uncertainties. Using the GPR time data from the moving wetting front only does not provide reliable results. Thus, we propose to use additional information from other types of reflectors to optimize the quality of the parameter estimation. Water movement and electromagnetic wave propagation in the unsaturated zone are modeled using a one-dimensional hydrogeophysical model. The GPR travel time data are analyzed for different reflectors: a moving reflector (the infiltration wetting front) and three fixed reflectors located at different depths in the soil. Global sensitivity analysis (GSA) is employed to assess the influence of the saturated hydraulic conductivity Ks, the saturated and residual water contents θs and θr, and the Mualem–van Genuchten shape parameters α and n of the soil on the GPR travel time data of the reflectors. Statistical calibration of the soil parameters is then performed using the MCMC method. The impact of the type of reflector (moving or fixed) is then evaluated by analyzing the calibrated model parameters and their confidence intervals for different scenarios. GSA results show that the sensitivities of the GPR data to the hydrodynamic soil parameters are different between moving and fixed reflectors, whereas fixed reflectors at various depths have similar sensitivities. Ks has a similar and strong influence on the data of both types of reflectors. Concerning the other parameters, for the wetting front, only θs and α have an influence, and only at long time steps since the total variance is zero at the very beginning of the experiment. On the other hand, for the fixed reflectors, the total variance is not zero at the very start and the parameters θs, θr, α and n can have an influence from the very beginning of the infiltration. Results of parameter estimation show that the use of calibration data from the moving or fixed reflectors alone does not enable a good identification of all soil parameters. With the moving reflector, the error between the estimated mean value and the exact target value for θr and α is 9 % and 45 %, respectively, and less than 3 % for the other parameters. The best reduction of the size of the parameter distribution is obtained for n, with a posterior distribution 9 times smaller than the prior one. For the others, this reduction ratio varies between 1 and 5. For the fixed reflectors, the estimated mean values are far from the target values for α, θr and n, representing for a reflector located at 120 cm 15 %, 27 %, and 121 %, respec","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"29 15","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-12-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138594127","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-12-06DOI: 10.5194/hess-27-4295-2023
C. Tschritter, C. Daughney, S. Karalliyadda, B. Hemmings, Uwe Morgenstern, C. Moore
Abstract. Groundwater age or residence time is important for identifying flow and contaminant pathways through groundwater systems. Typically, groundwater age and age distributions are inferred via lumped parameter models based on measured age tracer concentrations. However, due to cost and time constraints, age tracers are usually only sampled at a small percentage of the wells in a catchment. This paper describes and compares two methods to increase the number of groundwater age data points and assist with validating age distributions inferred from lumped parameter models. Two machine learning techniques with different strengths were applied to develop two independent metamodels that each aim to establish relationships between the hydrochemical parameters and the modelled groundwater age distributions in one test catchment. Ensemble medians from the best model realisations per age distribution percentile were used for comparison with the results from traditional lumped parameter models based on age tracers. Results show that both metamodelling techniques predict age distributions from hydrochemistry with good correspondence to traditional lumped parameter model (LPM)-derived age distributions. Therefore, these techniques can be used to assist with the interpretation of lumped parameter models where age tracers have been sampled, and they can also be applied to predict groundwater age distributions for wells in a similar hydrogeological regime that have hydrochemistry data available but no age tracer data.
{"title":"Estimation of groundwater age distributions from hydrochemistry: comparison of two metamodelling algorithms in the Heretaunga Plains aquifer system, New Zealand","authors":"C. Tschritter, C. Daughney, S. Karalliyadda, B. Hemmings, Uwe Morgenstern, C. Moore","doi":"10.5194/hess-27-4295-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4295-2023","url":null,"abstract":"Abstract. Groundwater age or residence time is important for identifying flow and contaminant pathways through groundwater systems. Typically, groundwater age and age distributions are inferred via lumped parameter models based on measured age tracer concentrations. However, due to cost and time constraints, age tracers are usually only sampled at a small percentage of the wells in a catchment. This paper describes and compares two methods to increase the number of groundwater age data points and assist with validating age distributions inferred from lumped parameter models. Two machine learning techniques with different strengths were applied to develop two independent metamodels that each aim to establish relationships between the hydrochemical parameters and the modelled groundwater age distributions in one test catchment. Ensemble medians from the best model realisations per age distribution percentile were used for comparison with the results from traditional lumped parameter models based on age tracers. Results show that both metamodelling techniques predict age distributions from hydrochemistry with good correspondence to traditional lumped parameter model (LPM)-derived age distributions. Therefore, these techniques can be used to assist with the interpretation of lumped parameter models where age tracers have been sampled, and they can also be applied to predict groundwater age distributions for wells in a similar hydrogeological regime that have hydrochemistry data available but no age tracer data.\u0000","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"20 3","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138596062","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract. Streamflow recession, shaped by hydrological processes, runoff dynamics, and catchment storage, is heavily influenced by landscape structure and rainstorm characteristics. However, our understanding of how recession relates to landscape structure and rainstorm characteristics remains inconsistent, with limited research examining their combined impact. This study examines this interplay in shaping recession responses upon 291 sets of recession parameters obtained through the decorrelation process. The data originate from 19 subtropical mountainous rivers and cover events with a wide spectrum of rainfall amounts. Key findings indicate that the recession coefficient (a) increases while the exponent (b) decreases with the L/G ratio (the median of ratios between flow-path length and gradient), suggesting that longer and gentler hillslopes facilitate flow accumulation and aquifer connectivity, ultimately reducing nonlinearity. Additionally, in large catchments, the exponent (b) increases with increasing rainfall due to greater landscape heterogeneity. Conversely, in small catchments, it declines with rainfall, indicating that these catchments have less landscape heterogeneity and thus reduced runoff heterogeneity. Our findings underscore the necessity for further validation of how L/G and drainage area regulate recession responses to varying rainfall levels across diverse regions.
{"title":"Landscape structures regulate the contrasting response of recession along rainfall amounts","authors":"Jun-Yi Lee, Ci-Jian Yang, Tsung-Ren Peng, Tsung-Yu Lee, Jr‐Chuan Huang","doi":"10.5194/hess-27-4279-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4279-2023","url":null,"abstract":"Abstract. Streamflow recession, shaped by hydrological processes, runoff dynamics, and catchment storage, is heavily influenced by landscape structure and rainstorm characteristics. However, our understanding of how recession relates to landscape structure and rainstorm characteristics remains inconsistent, with limited research examining their combined impact. This study examines this interplay in shaping recession responses upon 291 sets of recession parameters obtained through the decorrelation process. The data originate from 19 subtropical mountainous rivers and cover events with a wide spectrum of rainfall amounts. Key findings indicate that the recession coefficient (a) increases while the exponent (b) decreases with the L/G ratio (the median of ratios between flow-path length and gradient), suggesting that longer and gentler hillslopes facilitate flow accumulation and aquifer connectivity, ultimately reducing nonlinearity. Additionally, in large catchments, the exponent (b) increases with increasing rainfall due to greater landscape heterogeneity. Conversely, in small catchments, it declines with rainfall, indicating that these catchments have less landscape heterogeneity and thus reduced runoff heterogeneity. Our findings underscore the necessity for further validation of how L/G and drainage area regulate recession responses to varying rainfall levels across diverse regions.\u0000","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"27 1","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138596431","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-12-04DOI: 10.5194/hess-27-4257-2023
Samuel Morin, H. François, M. Réveillet, E. Sauquet, L. Crochemore, F. Branger, Étienne Leblois, Marie Dumont
Abstract. The presence of a ski resort modifies the snow cover at the local scale, due to snow management practices on ski pistes, especially grooming and snowmaking. Snow management exerts 2-fold effects on the local hydrological cycle, through (i) abstraction and transfer of water used for snowmaking, and (ii) changes in water runoff due to added snow mass through snowmaking and/or delayed melting of the snowpack due to snow grooming. This induces a local pressure on water resources, which has seldom been addressed in scientific studies hitherto. Here we introduce a method to compute the hydrological effects of snow management on ski pistes and we apply and illustrate its results for the case study of the La Plagne ski resort in the Northern French Alps. The approach mainly relies on snow cover modelling using the Crocus snow cover driven by SAFRAN reanalysis and climate projections. Model results are evaluated against in-situ hydrological observations and show that the modelling approach, although very simplified for many hydrological processes, provides relevant information and insights in terms of the influence of snow-related processes on water resources. Our study shows a visible impact of grooming, virtually eliminating snowmelt in winter, thus delaying the onset of snowmelt. This results is a lower snowmelt flux during the wintertime, low flow period, on the order of −10 % to −20 %, compensated by higher amounts when snow melts. While about 10 % of the water used for snowmaking is estimated to be lost by evaporation through the ice formation process from the liquid water droplets, we find that, in the case studied, the annual scale alteration of water resources is limited and estimated to be on the order of 1 % to 2 %. This is due to the fact that the amount of water used for snowmaking on ski pistes represents a fraction of 10 % to 20 % of total annual precipitation, that ski pistes cover typically 10 % of the surface area of catchments within which ski resorts are located, and that snowmaking equipment covers, in the case of La Plagne, 40 % of the surface area of ski pistes. Therefore, in this case, snowmaking mainly leads to a moderate shift in snow cover formation and snowmelt processes and plays, for example, a smaller role than the influence of future climate change on mountain hydrology. This study provides an initial overview of the influence of grooming and snowmaking on river flows in a mountain catchment, which can inform future studies on water management and climate change adaptation in areas with ski tourism facilities. This study does not discuss long-term sustainability challenges of ski tourism and other aspects of the local environmental impacts (landscape, biodiversity) of snow management, such as the construction and use of mountain water reservoirs and other earthworks in ski resorts.
{"title":"Simulated hydrological effects of grooming and snowmaking in a ski resort on the local water balance","authors":"Samuel Morin, H. François, M. Réveillet, E. Sauquet, L. Crochemore, F. Branger, Étienne Leblois, Marie Dumont","doi":"10.5194/hess-27-4257-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4257-2023","url":null,"abstract":"Abstract. The presence of a ski resort modifies the snow cover at the local scale, due to snow management practices on ski pistes, especially grooming and snowmaking. Snow management exerts 2-fold effects on the local hydrological cycle, through (i) abstraction and transfer of water used for snowmaking, and (ii) changes in water runoff due to added snow mass through snowmaking and/or delayed melting of the snowpack due to snow grooming. This induces a local pressure on water resources, which has seldom been addressed in scientific studies hitherto. Here we introduce a method to compute the hydrological effects of snow management on ski pistes and we apply and illustrate its results for the case study of the La Plagne ski resort in the Northern French Alps. The approach mainly relies on snow cover modelling using the Crocus snow cover driven by SAFRAN reanalysis and climate projections. Model results are evaluated against in-situ hydrological observations and show that the modelling approach, although very simplified for many hydrological processes, provides relevant information and insights in terms of the influence of snow-related processes on water resources. Our study shows a visible impact of grooming, virtually eliminating snowmelt in winter, thus delaying the onset of snowmelt. This results is a lower snowmelt flux during the wintertime, low flow period, on the order of −10 % to −20 %, compensated by higher amounts when snow melts. While about 10 % of the water used for snowmaking is estimated to be lost by evaporation through the ice formation process from the liquid water droplets, we find that, in the case studied, the annual scale alteration of water resources is limited and estimated to be on the order of 1 % to 2 %. This is due to the fact that the amount of water used for snowmaking on ski pistes represents a fraction of 10 % to 20 % of total annual precipitation, that ski pistes cover typically 10 % of the surface area of catchments within which ski resorts are located, and that snowmaking equipment covers, in the case of La Plagne, 40 % of the surface area of ski pistes. Therefore, in this case, snowmaking mainly leads to a moderate shift in snow cover formation and snowmelt processes and plays, for example, a smaller role than the influence of future climate change on mountain hydrology. This study provides an initial overview of the influence of grooming and snowmaking on river flows in a mountain catchment, which can inform future studies on water management and climate change adaptation in areas with ski tourism facilities. This study does not discuss long-term sustainability challenges of ski tourism and other aspects of the local environmental impacts (landscape, biodiversity) of snow management, such as the construction and use of mountain water reservoirs and other earthworks in ski resorts.\u0000","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"68 20","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-12-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138605046","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-12-01DOI: 10.5194/hess-27-4247-2023
Licong Dai, Ruiyu Fu, Xiaowei Guo, Yangong Du, G. Cao, Huakun Zhou, Zhongmin Hu
Abstract. Biocrust is a key component of ecosystems and plays a vital role in altering hydrological processes in terrestrial ecosystems. The impacts of biocrust on hydrological processes in arid and semi-arid ecosystems have been widely documented. However, the effects and mechanisms of biocrust on soil hydrological processes in alpine ecosystems are still poorly understood. In this study, we selected two meadow types from the northern Qinghai–Tibet Plateau: normal Kobresia meadow (NM) and biocrust meadow (BM). Both the soil hydrological and physicochemical properties were examined. We found that, in the 0–30 cm soil layer, soil water retention and soil water content in NM were higher than those in BM, whereas the 30–40 cm layer's soil water retention and soil water content in NM were lower than those in BM. The topsoil infiltration rate in BM was lower than that in NM. Furthermore, the physicochemical properties were different between NM and BM. The 0–10 cm soil layer's clay content in BM was 9 % higher than that in NM, whereas the 0–30 cm layer's soil capillary porosity in NM was higher than that in BM. In addition, the 0–20 cm layer's soil total nitrogen (TN) and soil organic matter (SOM) in NM were higher than those in BM, implying that the presence of biocrust may not favor the formation of soil nutrients owing to its lower soil microbial biomass carbon and microbial biomass nitrogen. Overall, soil water retention was determined by SOM by altering the soil capillary porosity and bulk density. Our findings suggest that the establishment of cyanobacteria crust biocrust may not improve soil water retention and infiltration, and the soil in cyanobacteria crust meadows could be more vulnerable to runoff generation and consequent soil erosion. These results provide a systematic and comprehensive understanding of the effects of biocrust on the soil hydrology of alpine ecosystems.
{"title":"Biocrust-reduced soil water retention and soil infiltration in an alpine Kobresia meadow","authors":"Licong Dai, Ruiyu Fu, Xiaowei Guo, Yangong Du, G. Cao, Huakun Zhou, Zhongmin Hu","doi":"10.5194/hess-27-4247-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4247-2023","url":null,"abstract":"Abstract. Biocrust is a key component of ecosystems and plays a vital role in altering hydrological processes in terrestrial ecosystems. The impacts of biocrust on hydrological processes in arid and semi-arid ecosystems have been widely documented. However, the effects and mechanisms of biocrust on soil hydrological processes in alpine ecosystems are still poorly understood. In this study, we selected two meadow types from the northern Qinghai–Tibet Plateau: normal Kobresia meadow (NM) and biocrust meadow (BM). Both the soil hydrological and physicochemical properties were examined. We found that, in the 0–30 cm soil layer, soil water retention and soil water content in NM were higher than those in BM, whereas the 30–40 cm layer's soil water retention and soil water content in NM were lower than those in BM. The topsoil infiltration rate in BM was lower than that in NM. Furthermore, the physicochemical properties were different between NM and BM. The 0–10 cm soil layer's clay content in BM was 9 % higher than that in NM, whereas the 0–30 cm layer's soil capillary porosity in NM was higher than that in BM. In addition, the 0–20 cm layer's soil total nitrogen (TN) and soil organic matter (SOM) in NM were higher than those in BM, implying that the presence of biocrust may not favor the formation of soil nutrients owing to its lower soil microbial biomass carbon and microbial biomass nitrogen. Overall, soil water retention was determined by SOM by altering the soil capillary porosity and bulk density. Our findings suggest that the establishment of cyanobacteria crust biocrust may not improve soil water retention and infiltration, and the soil in cyanobacteria crust meadows could be more vulnerable to runoff generation and consequent soil erosion. These results provide a systematic and comprehensive understanding of the effects of biocrust on the soil hydrology of alpine ecosystems.\u0000","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"69 3","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138623658","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-11-30DOI: 10.5194/hess-27-4227-2023
Roberto Bentivoglio, E. Isufi, S. Jonkman, Riccardo Taormina
Abstract. Numerical modelling is a reliable tool for flood simulations, but accurate solutions are computationally expensive. In recent years, researchers have explored data-driven methodologies based on neural networks to overcome this limitation. However, most models are only used for a specific case study and disregard the dynamic evolution of the flood wave. This limits their generalizability to topographies that the model was not trained on and in time-dependent applications. In this paper, we introduce shallow water equation–graph neural network (SWE–GNN), a hydraulics-inspired surrogate model based on GNNs that can be used for rapid spatio-temporal flood modelling. The model exploits the analogy between finite-volume methods used to solve SWEs and GNNs. For a computational mesh, we create a graph by considering finite-volume cells as nodes and adjacent cells as being connected by edges. The inputs are determined by the topographical properties of the domain and the initial hydraulic conditions. The GNN then determines how fluxes are exchanged between cells via a learned local function. We overcome the time-step constraints by stacking multiple GNN layers, which expand the considered space instead of increasing the time resolution. We also propose a multi-step-ahead loss function along with a curriculum learning strategy to improve the stability and performance. We validate this approach using a dataset of two-dimensional dike breach flood simulations in randomly generated digital elevation models generated with a high-fidelity numerical solver. The SWE–GNN model predicts the spatio-temporal evolution of the flood for unseen topographies with mean average errors in time of 0.04 m for water depths and 0.004 m2 s−1 for unit discharges. Moreover, it generalizes well to unseen breach locations, bigger domains, and longer periods of time compared to those of the training set, outperforming other deep-learning models. On top of this, SWE–GNN has a computational speed-up of up to 2 orders of magnitude faster than the numerical solver. Our framework opens the doors to a new approach to replace numerical solvers in time-sensitive applications with spatially dependent uncertainties.
{"title":"Rapid spatio-temporal flood modelling via hydraulics-based graph neural networks","authors":"Roberto Bentivoglio, E. Isufi, S. Jonkman, Riccardo Taormina","doi":"10.5194/hess-27-4227-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4227-2023","url":null,"abstract":"Abstract. Numerical modelling is a reliable tool for flood simulations, but accurate solutions are computationally expensive. In recent years, researchers have explored data-driven methodologies based on neural networks to overcome this limitation. However, most models are only used for a specific case study and disregard the dynamic evolution of the flood wave. This limits their generalizability to topographies that the model was not trained on and in time-dependent applications. In this paper, we introduce shallow water equation–graph neural network (SWE–GNN), a hydraulics-inspired surrogate model based on GNNs that can be used for rapid spatio-temporal flood modelling. The model exploits the analogy between finite-volume methods used to solve SWEs and GNNs. For a computational mesh, we create a graph by considering finite-volume cells as nodes and adjacent cells as being connected by edges. The inputs are determined by the topographical properties of the domain and the initial hydraulic conditions. The GNN then determines how fluxes are exchanged between cells via a learned local function. We overcome the time-step constraints by stacking multiple GNN layers, which expand the considered space instead of increasing the time resolution. We also propose a multi-step-ahead loss function along with a curriculum learning strategy to improve the stability and performance. We validate this approach using a dataset of two-dimensional dike breach flood simulations in randomly generated digital elevation models generated with a high-fidelity numerical solver. The SWE–GNN model predicts the spatio-temporal evolution of the flood for unseen topographies with mean average errors in time of 0.04 m for water depths and 0.004 m2 s−1 for unit discharges. Moreover, it generalizes well to unseen breach locations, bigger domains, and longer periods of time compared to those of the training set, outperforming other deep-learning models. On top of this, SWE–GNN has a computational speed-up of up to 2 orders of magnitude faster than the numerical solver. Our framework opens the doors to a new approach to replace numerical solvers in time-sensitive applications with spatially dependent uncertainties.","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"291 1","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-11-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139199937","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-11-28DOI: 10.5194/hess-27-4217-2023
Germano G. Ribeiro Neto, Sarra Kchouk, L. Melsen, L. Cavalcante, D. W. Walker, A. Dewulf, A. Costa, E. S. P. R. Martins, P. V. van Oel
Abstract. Human actions induce and modify droughts. However, scientific gaps remain with respect to how hydrological processes, anthropogenic dynamics, and individuals' perceptions of impacts are intrinsically entangled in drought occurrence and evolution. This adds complexity to drought assessment studies that cannot be addressed by the natural and environmental sciences alone. Furthermore, it poses a challenge with respect to developing ways to evaluate human behaviour and its pattern of co-evolution with the hydrological cycle – mainly related to water use and landscape modifications. During fieldwork in Brazil, we observed how drought impacts were experienced by people who were exposed to a multi-year drought. Evaluating our data, it appeared that prospect theory, a behavioural economic theory that is usually applied to explain decision-making processes under uncertainty, has explanatory power regarding what we observed in the field. Therefore, we propose an interdisciplinary approach to improve the understanding of drought impact emergence using this theory. When employing prospect theory in this context, drought impacts are considered failed welfare expectations (“prospects”) due to water shortage. A shifting baseline after prolonged exposure to drought can therefore mitigate experienced drought impacts. We demonstrate that this theory can also contribute to explaining socio-hydrological phenomena, such as reservoir effects. This new approach can help bridge natural science and social science perspectives, resulting in integrated drought management that considers the local context.
{"title":"HESS Opinions: Drought impacts as failed prospects","authors":"Germano G. Ribeiro Neto, Sarra Kchouk, L. Melsen, L. Cavalcante, D. W. Walker, A. Dewulf, A. Costa, E. S. P. R. Martins, P. V. van Oel","doi":"10.5194/hess-27-4217-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4217-2023","url":null,"abstract":"Abstract. Human actions induce and modify droughts. However, scientific gaps remain with respect to how hydrological processes, anthropogenic dynamics, and individuals' perceptions of impacts are intrinsically entangled in drought occurrence and evolution. This adds complexity to drought assessment studies that cannot be addressed by the natural and environmental sciences alone. Furthermore, it poses a challenge with respect to developing ways to evaluate human behaviour and its pattern of co-evolution with the hydrological cycle – mainly related to water use and landscape modifications. During fieldwork in Brazil, we observed how drought impacts were experienced by people who were exposed to a multi-year drought. Evaluating our data, it appeared that prospect theory, a behavioural economic theory that is usually applied to explain decision-making processes under uncertainty, has explanatory power regarding what we observed in the field. Therefore, we propose an interdisciplinary approach to improve the understanding of drought impact emergence using this theory. When employing prospect theory in this context, drought impacts are considered failed welfare expectations (“prospects”) due to water shortage. A shifting baseline after prolonged exposure to drought can therefore mitigate experienced drought impacts. We demonstrate that this theory can also contribute to explaining socio-hydrological phenomena, such as reservoir effects. This new approach can help bridge natural science and social science perspectives, resulting in integrated drought management that considers the local context.","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"14 1","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139225795","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-11-28DOI: 10.5194/hess-27-4205-2023
Chloé Fandel, T. Ferré, François Miville, P. Renard, N. Goldscheider
Abstract. Reconstructing the geologic history of a karst area can advance understanding of the system's present-day hydrogeologic functioning and help predict the location of unexplored conduits. This study tests competing hypotheses describing past conditions controlling cave formation in an alpine karst catchment, by comparing an ensemble of modeled networks to the observed network map. The catchment, the Gottesacker karst system (Germany and Austria), is drained by three major springs and a paleo-spring and includes the partially explored Hölloch cave, which consists of an active section whose formation is well-understood and an inactive section whose formation is the subject of debate. Two hypotheses for the formation of the inactive section are the following: (1) glaciation obscured the three present-day springs, leaving only the paleo-spring, or (2) the lowest of the three major springs (Sägebach) is comparatively young, so its subcatchment previously drained to the paleo-spring. These hypotheses were tested using the pyKasso Python library (built on anisotropic fast-marching methods) to generate two ensembles of networks, one representing each scenario. Each ensemble was then compared to the known cave map. The simulated networks generated under hypothesis 2 match the observed cave map more closely than those generated under hypothesis 1. This supports the conclusion that the Sägebach spring is young, and it suggests that the cave likely continues southwards. Finally, this study extends the applicability of model ensemble methods from situations where the geologic setting is known but the network is unknown to situations where the network is known but the geologic evolution is not.
{"title":"Improving understanding of groundwater flow in an alpine karst system by reconstructing its geologic history using conduit network model ensembles","authors":"Chloé Fandel, T. Ferré, François Miville, P. Renard, N. Goldscheider","doi":"10.5194/hess-27-4205-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4205-2023","url":null,"abstract":"Abstract. Reconstructing the geologic history of a karst area can advance understanding of the system's present-day hydrogeologic functioning and help predict the location of unexplored conduits. This study tests competing hypotheses describing past conditions controlling cave formation in an alpine karst catchment, by comparing an ensemble of modeled networks to the observed network map. The catchment, the Gottesacker karst system (Germany and Austria), is drained by three major springs and a paleo-spring and includes the partially explored Hölloch cave, which consists of an active section whose formation is well-understood and an inactive section whose formation is the subject of debate. Two hypotheses for the formation of the inactive section are the following: (1) glaciation obscured the three present-day springs, leaving only the paleo-spring, or (2) the lowest of the three major springs (Sägebach) is comparatively young, so its subcatchment previously drained to the paleo-spring. These hypotheses were tested using the pyKasso Python library (built on anisotropic fast-marching methods) to generate two ensembles of networks, one representing each scenario. Each ensemble was then compared to the known cave map. The simulated networks generated under hypothesis 2 match the observed cave map more closely than those generated under hypothesis 1. This supports the conclusion that the Sägebach spring is young, and it suggests that the cave likely continues southwards. Finally, this study extends the applicability of model ensemble methods from situations where the geologic setting is known but the network is unknown to situations where the network is known but the geologic evolution is not.","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"27 1","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139222732","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-11-21DOI: 10.5194/hess-27-4187-2023
Yuan Li, Kangning Xü, Zhiyong Wu, Zhiwei Zhu, Quan J. Wang
Abstract. In this study, we develop a spatial–temporal projection-based calibration, bridging, and merging (STP-CBaM) method to improve probabilistic sub-seasonal precipitation forecast skill over 17 hydroclimatic regions in China. The calibration model is established by post-processing ECMWF raw forecasts using the Bayesian joint probability (BJP) approach. The bridging models are built using large-scale atmospheric intraseasonal predictors, including zonal wind at 200 hPa (U200) and 850 hPa (U850); an outgoing longwave radiation anomaly (OLRA); and geopotential height at 200 hPa (H200), 500 hPa (H500), and 850 hPa (H850) defined by the STP method. The calibration model and the bridging models are then merged through the Bayesian modelling averaging (BMA) method. Our results indicate that the forecast skill of the calibration model is higher compared to bridging models when the lead time is within 5–10 d. The U200- and OLRA-based bridging models outperform the calibration model in certain months and certain regions. The BMA-merged forecasts take advantage of both calibration models and bridging models. Meanwhile, the BMA-merged forecasts also show high reliability at longer lead times. However, some improvements to reliability are still needed at shorter lead times. These findings demonstrate the great potential to combine dynamical models and statistical models in improving sub-seasonal precipitation forecasts.
{"title":"A statistical–dynamical approach for probabilistic prediction of sub-seasonal precipitation anomalies over 17 hydroclimatic regions in China","authors":"Yuan Li, Kangning Xü, Zhiyong Wu, Zhiwei Zhu, Quan J. Wang","doi":"10.5194/hess-27-4187-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4187-2023","url":null,"abstract":"Abstract. In this study, we develop a spatial–temporal projection-based calibration, bridging, and merging (STP-CBaM) method to improve probabilistic sub-seasonal precipitation forecast skill over 17 hydroclimatic regions in China. The calibration model is established by post-processing ECMWF raw forecasts using the Bayesian joint probability (BJP) approach. The bridging models are built using large-scale atmospheric intraseasonal predictors, including zonal wind at 200 hPa (U200) and 850 hPa (U850); an outgoing longwave radiation anomaly (OLRA); and geopotential height at 200 hPa (H200), 500 hPa (H500), and 850 hPa (H850) defined by the STP method. The calibration model and the bridging models are then merged through the Bayesian modelling averaging (BMA) method. Our results indicate that the forecast skill of the calibration model is higher compared to bridging models when the lead time is within 5–10 d. The U200- and OLRA-based bridging models outperform the calibration model in certain months and certain regions. The BMA-merged forecasts take advantage of both calibration models and bridging models. Meanwhile, the BMA-merged forecasts also show high reliability at longer lead times. However, some improvements to reliability are still needed at shorter lead times. These findings demonstrate the great potential to combine dynamical models and statistical models in improving sub-seasonal precipitation forecasts.","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"17 11","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139251349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"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 : 2023-11-21DOI: 10.5194/hess-27-4173-2023
Emily I. Burt, Gregory R. Goldsmith, Roxanne M. Cruz-de Hoyos, Adan J. Ccahuana Quispe, A. J. West
Abstract. Determining the sources of water provisioning streams, soils, and vegetation can provide important insights into the water that sustains critical ecosystem functions now and how those functions may be expected to respond given projected changes in the global hydrologic cycle. We developed multi-year time series of water isotope ratios (δ18O and δ2H) based on twice-monthly collections of precipitation, lysimeter, and tree branch xylem waters from a seasonally dry tropical montane cloud forest in the southeastern Andes mountains of Peru. We then used this information to determine indices of the seasonal origins, the young water fractions (Fyw), and the new water fractions (Fnew) of soil, stream, and tree water. There was no evidence for intra-annual variation in the seasonal origins of stream water and lysimeter water from 1 m depth, both of which were predominantly comprised of wet-season precipitation even during the dry seasons. However, branch xylem waters demonstrated an intra-annual shift in seasonal origin: xylem waters were comprised of wet-season precipitation during the wet season and dry-season precipitation during the dry season. The young water fractions of lysimeter (< 15 %) and stream (5 %) waters were lower than the young water fraction (37 %) in branch xylem waters. The new water fraction (an indicator of water ≤ 2 weeks old in this study) was estimated to be 12 % for branch xylem waters, while there was no significant evidence for new water in stream or lysimeter waters from 1 m depth. Our results indicate that the source of water for trees in this system varied seasonally, such that recent precipitation may be more immediately taken up by shallow tree roots. In comparison, the source of water for soils and streams did not vary seasonally, such that precipitation may mix and reside in soils and take longer to transit into the stream. Our insights into the seasonal origins and ages of water in soils, streams, and vegetation in this humid tropical montane cloud forest add to understanding of the mechanisms that govern the partitioning of water moving through different ecosystems.
{"title":"The seasonal origins and ages of water provisioning streams and trees in a tropical montane cloud forest","authors":"Emily I. Burt, Gregory R. Goldsmith, Roxanne M. Cruz-de Hoyos, Adan J. Ccahuana Quispe, A. J. West","doi":"10.5194/hess-27-4173-2023","DOIUrl":"https://doi.org/10.5194/hess-27-4173-2023","url":null,"abstract":"Abstract. Determining the sources of water provisioning streams, soils, and vegetation can provide important insights into the water that sustains critical ecosystem functions now and how those functions may be expected to respond given projected changes in the global hydrologic cycle. We developed multi-year time series of water isotope ratios (δ18O and δ2H) based on twice-monthly collections of precipitation, lysimeter, and tree branch xylem waters from a seasonally dry tropical montane cloud forest in the southeastern Andes mountains of Peru. We then used this information to determine indices of the seasonal origins, the young water fractions (Fyw), and the new water fractions (Fnew) of soil, stream, and tree water. There was no evidence for intra-annual variation in the seasonal origins of stream water and lysimeter water from 1 m depth, both of which were predominantly comprised of wet-season precipitation even during the dry seasons. However, branch xylem waters demonstrated an intra-annual shift in seasonal origin: xylem waters were comprised of wet-season precipitation during the wet season and dry-season precipitation during the dry season. The young water fractions of lysimeter (< 15 %) and stream (5 %) waters were lower than the young water fraction (37 %) in branch xylem waters. The new water fraction (an indicator of water ≤ 2 weeks old in this study) was estimated to be 12 % for branch xylem waters, while there was no significant evidence for new water in stream or lysimeter waters from 1 m depth. Our results indicate that the source of water for trees in this system varied seasonally, such that recent precipitation may be more immediately taken up by shallow tree roots. In comparison, the source of water for soils and streams did not vary seasonally, such that precipitation may mix and reside in soils and take longer to transit into the stream. Our insights into the seasonal origins and ages of water in soils, streams, and vegetation in this humid tropical montane cloud forest add to understanding of the mechanisms that govern the partitioning of water moving through different ecosystems.","PeriodicalId":13143,"journal":{"name":"Hydrology and Earth System Sciences","volume":"65 11","pages":""},"PeriodicalIF":6.3,"publicationDate":"2023-11-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139254297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：