Abstract. Hydrological forecasts are important for operational water management and near future planning, even more so in light of increased occurrences of extreme events such as floods and droughts. Having a flexible forecasting framework that can deliver this information in fast and computational efficient manner is critical. In this study, the suitability of a hybrid forecasting framework, combining data-driven approaches and seasonal (re)forecasting information to predict hydrological variables was explored. Target variables include discharge and surface water levels for various stations at national scale with the Netherlands as focus. Five different ML models, ranging from simple to more complex and trained on historical observations of discharge, precipitation, evaporation and sea water levels, were run with seasonal (re)forecast data (EFAS and SEAS5) of these driver variables in a hindcast setting. The results were evaluated using the evaluation metrics Anomaly Correlation Coefficient (ACC), Continuous Ranked Probability (Skill) Score (CRPS and CRPSS), and Brier Skill Score (BSS) in comparison to a climatological reference hindcast. Aggregating results of all stations and ML models revealed that the hindcasting framework outperformed the climatological reference forecasts by roughly 60 % for discharge predictions (80 % for surface water level predictions). Skilful prediction for the first lead month, independently of initialization month, can be made for discharge. The skill extends up to 2–3 months for spring months due to snow melt dynamics captured in the training phase of the model. Surface water levels hindcasts showed similar skill and skilful lead times. While the different ML models showed differences in performance during a testing and training phase using historical observations, running the ML framework in a hindcast setting showed only minor differences between the models, which is attributed to the uncertainty in seasonal forecasts. However, despite being trained on historical observations, the hybrid framework used in this study shows similar skilful predictions as previous large scale forecasting systems. With our study we show that a hybrid framework is able to bring location specific skilful seasonal forecast information with global seasonal forecast inputs. At the same time our hybrid approach is flexible and fast, and as such a hybrid framework could be adapted to make it even more interesting to water managers and their needs, for instance a part of a fast model-predictive control framework.
Abstract. Hydrological forecasts are important for operational water management and near-future planning, even more so in light of the increased occurrences of extreme events such as floods and droughts. Having a forecasting framework, which is flexible in terms of input forcings and forecasting locations (local, regional, or national) that can deliver this information in fast and computational efficient manner, is critical. In this study, the suitability of a hybrid forecasting framework, combining data-driven approaches and seasonal (re)forecasting information from dynamical models, to predict hydrological variables was explored. Target variables include discharge and surface water levels for various stations at a national scale, with the Netherlands as the focus. Five different machine learning (ML) models, ranging from simple to more complex and trained on historical observations of discharge, precipitation, evaporation, and seawater levels, were run with seasonal (re)forecast data, including the European Flood Awareness System (EFAS) and ECMWF seasonal forecast system (SEAS5), of these driver variables in a hindcast setting. The results were evaluated using the evaluation metrics, i.e. anomaly correlation coefficient (ACC), continuous ranked probability (skill) score (CRPS and CRPSS), and Brier skill score (BSS), in comparison to a climatological reference hindcast. Aggregating the results of all stations and ML models revealed that the hindcasting framework outperformed the climatological reference forecasts by roughly 60 % for discharge predictions (80 % for surface water level predictions). Skilful prediction for the first lead month, independently of the initialization month, can be made for discharge. The skill extends up to 2–3 months for spring months due to snowmelt dynamic captured in the training phase of the model. Surface water level hindcasts showed similar skill and skilful lead times. While the different ML models showed differences in performance during a testing and training phase using historical observations, running the ML framework in a hindcast setting showed only minor differences between the models, which is attributed to the uncertainty in seasonal forecasts. However, despite being trained on historical observations, the hybrid framework used in this study shows similar skilful predictions to previous large-scale forecasting systems. With our study, we show that a hybrid framework is able to bring location-specific skilful seasonal forecast information with global seasonal forecast inputs. At the same time, our hybrid approach is flexible and fast, and as such, a hybrid framework could be adapted to make it even more interesting to water managers and their needs, for instance, as part of a fast model-predictive control framework.
<p>Climate change has a large influence on the occurrence of extreme hydrological events. In this study we take advantage of two recent developments that allow for more detailed and local estimates of future hydrological extremes. New large climate ensembles (LE) now provide more insight into the occurrence of hydrological extremes as they offer order of magnitude more realizations of the future weather and thus floods and droughts. At the same time recent developments in Machine Learning (ML) based forecasting have enabled scientists to provide this LE information to a local scale relevant to water managers.</p> <p>In this study we combine LE, consisting of 2000 years of global data for scenarios representing present-day, 2 and 3 degrees warmer climate <sup>(1)</sup>, together with a local, observation-based ML model framework for simulating hydrological extremes for the Netherlands <sup>(2, 3)</sup>.</p> <p>We developed a new post-processing approach that allows us to use LE simulation data for local applications based on historical information. We test the application of the post-processing step based on historical simulations, before implementing in the different scenario runs.</p> <p>The discharge simulation results for the different scenarios show a clear seasonal cycle with increased low flow periods (both average duration and number of events) from summer till end of autumn (~45% August-October) and increased high flow periods for early spring (~43% February-April) looking at national scale, with the 3-degree warmer climate scenario showing the highest percentages for both (52.5% and 48.3% respectively). Regional differences can be seen in terms of shifts (low flows occurring earlier in the year) and range (higher/lower percentages). These trends can further be detangled into location specific results, due to the added value provided by the ML setup.</p> <p>We show that by combining the wealth of information from LE and the speed and accuracy of ML models we can advance the state-of-the-art when it comes to modelling and projecting hydrological extremes. The local modelling framework allows to simulate discharge under different climate change scenarios for national, regional and local scale assessments. The historically and locally trained models provide essential information for water management to be used in&#160; long-term planning.</p> <p><em><sup>1)</sup></em><em> Van der Wiel, K., Wanders, N., Selten, F. M., & Bierkens, M. F. P. (2019). </em><em>Added value of large ensemble simulations for assessing extreme river discharge in a 2 &#176;C warmer world. </em><em>Geophysical Research Letters, 46, 2093&#8211; 2102. <br /><sup>2)</sup> Hauswirth, S. M., Bierkens, M. F., Beijk, V., & Wanders, N. (2021). </em><em>The potential of data driven approaches for quantifying hydrological extremes. Advances in Water Resources, 155, 104017.<br /></em><em><sup>3)</sup></em><em> Hauswirth, S. M., Bierkens, M. F., Beijk, V., & Wanders, N. (2022). </em><em>The suitability of a hybrid framework including data driven approaches for hydrological forecasting. Hydrology and Earth System Sciences Discussions, 1-20.</em></p>
IntroductionClimate change has a large influence on the occurrence of extreme hydrological events. However, reliable estimates of future extreme event probabilities, especially when needed locally, require very long time series with hydrological models, which is often not possible due to computational constraints. In this study we take advantage of two recent developments that allow for more detailed and local estimates of future hydrological extremes. New large climate ensembles (LE) now provide more insight on the occurrence of hydrological extremes as they offer order of magnitude more realizations of future weather. At the same time recent developments in Machine Learning (ML) in hydrology create great opportunities to study current and upcoming problems in a new way, including and combining large amounts of data.MethodsIn this study, we combined LE together with a local, observation based ML model framework with the goal to see if and how these aspects can be combined and to simulate, assess and produce estimates of hydrological extremes under different warming levels for local scales. For this, first a new post-processing approach was developed that allowed us to use LE simulation data for local applications. The simulation results of discharge extreme events under different warming levels were assessed in terms of frequency, duration and intensity and number of events at national, regional and local scales.ResultsClear seasonal cycles with increased low flow frequency were observed for summer and autumn months as well as increased high flow periods for early spring. For both extreme events, the 3C warmer climate scenario showed the highest percentages. Regional differences were seen in terms of shifts and range. These trends were further refined into location specific results. The shifts and trends observed between the different scenarios were due to a change in climate variability.DiscussionIn this study we show that by combining the wealth of information from LE and the speed and local relevance of ML models we can advance the state-of-the-art when it comes to modeling hydrological extremes under different climate change scenarios for national, regional and local scale assessments providing relevant information for water management in terms of long term planning.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.