Orals: Thu, 1 May | Room 2.15
Despite advancements in science and technology, predicting and preparing for floods remains challenging due to uncertainties in forecasting the atmospheric and hydrologic processes, limited real-time data, and forecast communication barriers. The Integrating Prediction of Precipitation and Hydrology for Early Actions (InPRHA) project is a five-year initiative within the World Meteorological Organization's (WMO) World Weather Research Programme (WWRP) that was initiated in 2024. It aims to enhance international research collaboration and scientific knowledge to improve flood hazard predictability and communication strategies for early warnings. The project integrates physical sciences, social sciences, and practitioner perspectives to advance hydro-meteorological forecasting and warning systems in a rapidly changing world.
Here, we present and reflect on the key scientific questions across the seven themes that constitute the implementation plan of the project, which embraces the broader research and operational communities. We focus on the following activities that guide the project’s implementation plan: DEFINE (identifying challenges), CONSTRUCT (gathering case studies), EXPERIMENT (scientific evaluations), and ENGAGE (community collaboration). These activities should bring people with shared interests together to drive transdisciplinary research and enhance flood forecasting systems worldwide for improved early action and decision-making. They call the community to address research needs on flood multi-hazard interdependencies, local vulnerability assessment and response, and climate change impacts on precipitation and hydrological forecasts. International cooperation is key to collaborate on addressing the current challenges of re-envisioning the warning process.
How to cite: Ramos, M.-H., Cattoen-Gilbert, C., and Hogan Carr, R. and the The InPRHA Steering Group: Integrating Prediction of Precipitation and Hydrology for Early Actions (InPRHA): what is still needed to ensure effective delivery and use of probabilistic hydro-meteorological forecasts from minutes to days ahead?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7983, https://doi.org/10.5194/egusphere-egu25-7983, 2025.
While Denmark has a long history of coastal floods and a robust storm flood warning system, forecasting fluvial flooding is a new challenge for the country presented by climate change. In 2022 the Danish Meteorological Institute (DMI) was assigned the responsibility for issuing national flood warnings in Denmark. DMI is mandated to send public flood warnings based on an ensemble of hydrological forecasting models from summer 2025.
This work presents the Danish experience in defining combined model ensemble criteria and warning levels approach for issuing flood warnings.
This study evaluates the criteria for determining the most reliable flood warnings using variants of two physically based classical hydrological models and one data-driven machine learning model. The models included in the ensemble are: (a) a physically based flood forecasting model based on the Hydrological Predictions for the Environment (HYPE) model (b) a variant of the HYPE model incorporating data assimilation, (c) a physically based MIKE-SHE model with detailed representation of groundwater, (d) a data-driven Long Short-Term Memory (LSTM) machine learning model using catchment characteristics and hydrological variables, (d) a hybrid model coupling of the LSTM model to the HYPE model.
While the HYPE and LSTM models have been developed by DMI, the MIKE-SHE model is developed by the Geological Survey of Denmark and Greenland (GEUS).
The performance of the ensemble model criteria is evaluated over the period 2011-2022 for warning stations across Denmark. We define warning levels based on extreme value statistics for observed discharge data to identify several return periods, issuing warnings when forecasted flows exceed these thresholds.
The study assesses the ensemble-based criterion that offers the best performance for issuing flood warnings across Denmark. We evaluate the performance using metrics such as Critical Success Index (CSI) and the Equitable Threat Score (ETS). Additionally, we examine trade-offs between the Success Rate and False Alarms and analyze spatial trends in model performance. We also reflect on the ease-of-use, scalability across Denmark and efficiency of the warning criteria because the criteria will ultimately be adopted for operational flood warning in real time in Denmark.
We find that it is important to consider the interplay between limitations in individual models in our ensemble and the choice of warning criteria to select a combination that provides a robust basis for issuing useful flood warnings.
Our experience in implementing Denmark's first flood warning system combining ensemble models with warning criteria offers valuable insights for countries where flooding is emerging as a new challenge brought by climate change.
How to cite: Payet-Burin, R., Henry Madsen, M., Thrysøe, C., Agata Plum, C., Dybro Thomassen, E., Martinsen, G., Wied Pedersen, J., Butts, M., Aarestrup, P., and Dhaubanjar, S.: Issuing flood warnings in Denmark based on an ensemble of hydrological forecast models., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-10471, https://doi.org/10.5194/egusphere-egu25-10471, 2025.
In recent years, small-scale heavy precipitation in Saxony has repeatedly led to exceptional flood situations in small and very small catchment areas in the low mountain range. Such catchment areas often react very quickly to precipitation events. Due to the sometimes tremendous effects, particularly in residential areas, early flood warning is essential. The forecast and the associated flood warning enable local authorities, emergency services and water authorities to prepare for the potential flood situation at an early stage and initiate the necessary measures.
We developed a flood early warning system that is in operational use in three pilot regions in Saxony. Data processing is event-driven and is controlled by a so-called sentinel component based on meteorological forecasts. This sentinel checks every 30 minutes whether a specific precipitation threshold in the forecast for the next 24 or 48 hours will be exceeded at any location in Saxony. In this case, the necessary precipitation data for a hydrological ensemble forecast is compiled. Two demonstrators were implemented for this purpose: (1) use of precipitation for observation and forecasting from established products of the German Weather Service (RADOLAN-RW – radar based QPE, RADOLAN-RV – radar based nowcasting, ICON-D2-EPS – ensemble QPF) and (2) precipitation from new, prototype products of the German Weather Service (pyRADMAN, SINFONY-INTENSE). With the pyRADMAN product, radar calibration is carried out by assimilation to rain gauges and commercial microwave links with a temporal resolution of 15 minutes. The SINFONY-INTENSE product integrates nowcasting and forecasts data into a seamless prediction ensemble with 21 members. Based on the respective combination of data, a separate decision is made for each warning region in Saxony as to whether a specific flood situation is imminent. If the system recognises a potential flood situation, the catchments in the according warning region are simulated using the event-based hydrological model DeHM. DeHM includes processes for runoff formation, runoff concentration, routing and simulation of dams. Data assimilation is carried out using online coupling with runoff data at gauging stations and reservoir levels in flood retention basins or dams.
The demonstrator with established products has been running since August 2023 and the demonstrator with prototype products since December 2023. The performance of both systems is evaluated. Parameters such as KGE or Percentage Error in Peak alongside threshold-based parameters such as False Alarm Ratio or Area under ROC Curve (AUC) allow the quality of both demonstrators to be assessed using various prediction ranges. The differences between the two demonstrators are shown on the basis of the quality measures and specific simulation results, and the associated benefits for early flood warning are discussed.
How to cite: Wagner, M. and Grundmann, J.: On the benefits of new, seamless prediction products in operational hydro-meteorological ensemble forecasting in small catchments, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15923, https://doi.org/10.5194/egusphere-egu25-15923, 2025.
Implementing flood forecasting models and early warning systems in decision-making can substantially reduce the impacts of floods and other natural hazards and provide more time for warnings and anticipatory action. Verification of flood forecasts is critical in providing emergency managers with practical information about the forecasts’ performance. However, incorporating forecasts in decision-making and warning applications demands a user-relevant categorical verification approach that covers practical aspects of operational systems. In this regard, standard scientific practice for verification may not match the forecast users’ experience or be suitable for all end users. It creates several challenges in categorical verification, including the vague definitions of the 2×2 contingency table categories, as they do not consider the slightly misaligned timing of events. Another challenge is the counting method of the categories, where different counting methods, based on a regular time-step basis, or based on individual events, may lead to different conclusions, and therefore, deliver different performance insights and value for forecast users. This research addresses these challenges, proposing a user-relevant verification approach, and examining the corresponding effects on the performance quality and economic value of flood forecasts. Two global forecasting models –Google AI model and GloFAS– are verified with the proposed approach in the African continent. Our results show a consistent pattern of decrease in performance when considering a user-relevant approach in comparison with standard verification practice. These findings emphasize a gap between standard and user-tailored verification, and the need for user-relevant verification of other flood forecasting systems, with the consideration of implementing additional aspects of operational systems.
How to cite: Alwahshat, M., Bertini, C., van Andel, S. J., and Nearing, G.: A user-relevant approach to the verification of flood forecasts, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-863, https://doi.org/10.5194/egusphere-egu25-863, 2025.
Seasonal hydrometeorological forecasts, often incorporated in a climate service, may provide (probabilistic) predictions or outlooks of water resources availability and early warning of drought to agriculture, hydropower, humanitarian, tourism, forestry, and other sectors. Availability of seasonal forecast data and of climate services have mushroomed in recent decades, but research shows that the actual uptake and use of these has not quite kept pace. Several reasons for this lag are identified, and it is acknowledged that three key dimensions that foster uptake are the credibility, salience and legitimacy of data and services provided; from the perspective of intended users.
As showcased in many contributions in this session, significant effort is dedicated to improving the consistency and quality of (probabilistic) seasonal forecasts through the understanding of uncertainty, data-assimilation, and bias-reduction. This is important, as an accurate forecast contributes to this being considered good, and the opinion users hold on its credibility. However, accuracy alone is not sufficient. Users must also consider the forecast salient, or relevant, to the decisions it is intended to support. Even the most precise and technically robust forecasts may fail to be considered if they do not align with the specific needs, priorities and contexts of users. This eludes to the title of this contribution, which is the dictionary definition of the word value. If users consider the predictions or outlooks actionable in informing their decision-making processes, then these hold value to them. Conversely, researchers and developers of seasonal hydrometeorological forecasts often find other aspects important. They may value forecasts that provide accurate predictions of hydrometeorological variables, and use forecast verification through a range of metrics to evaluate forecast goodness. While these metrics are essential from a technical perspective, they may be less meaningful to users.
In this contribution we explore how seasonal forecasts can be co-evaluated, together with users, and through user-oriented (verification) metrics. An essential step is the identification of needs through a common understanding of the decisions that users take and how they use data and knowledge in making those decisions. We show how this has been developed with users in Living Labs established in the I-CISK project, an EU research initiative. We find that this is a highly iterative process, with tools such as interviews, surveys, focus group discussions and co-developed decision timelines giving rich insight to what decisions are made, thresholds that are used, and when decisions are made when these vary seasonally. We also evaluate through interviews with users a selection of commonly used verification metrics. Results from these interviews show that decision-oriented metrics such as contingency tables are considered more informative than other metrics, as are visual inspection of forecast performance for past events. They also show that discussing these in the co-evaluation process contributed to the opinion users had on the credibility of the forecasts and how these are of value to them. Significantly, the co-evaluation of forecasts was also found to help build trust, contributing to legitimacy; the third important dimension of uptake.
How to cite: Werner, M., Dotta Correa, D., Egan, K., Baugh, C., van Andel, S.-J., and Emerton, R.: Value [def.]: The importance or worth of something for someone, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18976, https://doi.org/10.5194/egusphere-egu25-18976, 2025.
Forecasting river discharge is essential for disaster risk reduction and water resource management, but forecasts of future river states often contain errors. Post-processing reduces forecast errors but is usually only applied at the locations of river gauges, leaving the majority of the river network uncorrected. Here, we present a data-assimilation-inspired method for error-correcting ensemble simulations across gauged and ungauged locations in a post-processing step. Our new method employs state augmentation within the framework of the Localised Ensemble Transform Kalman Filter (LETKF) to estimate an error vector for each ensemble member. The LETKF uses ensemble error covariances to spread observational information from gauged to ungauged locations in a dynamic and computationally efficient manner. To improve the efficiency of the LETKF we define localisation, covariance inflation, and initial ensemble generation techniques that can be easily transferred between modelling systems and river catchments. We implement and evaluate our new error-correction method for the Rhine-Meuse catchment using ensemble forecasts from the Copernicus Emergency Management Service’s European Flood Awareness System (EFAS). The resulting river discharge ensembles are error-corrected at every grid box but remain spatially and temporally consistent. Leave-one-out cross validation is used to evaluate the skill of the ensembles at proxy-ungauged locations to assess the ability of the method to spread the correction along the river network. The skill of the ensemble mean is improved at almost all locations including stations both up- and downstream of the assimilated observations. Whilst the ensemble spread is improved at short lead-times, at longer lead-times the ensemble reliability is decreased. In summary, our method successfully propagates error information along the river network, enabling error correction at ungauged locations. This technique can be used for improved post-event analysis and can be developed further to post-process operational forecasts providing more accurate knowledge about the future states of rivers.
How to cite: Matthews, G., Cloke, H. L., Dance, S. L., and Prudhomme, C.: Error-correction across gauged and ungauged locations: A data assimilation-inspired approach to post-processing river discharge ensembles, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11733, https://doi.org/10.5194/egusphere-egu25-11733, 2025.
Seasonal forecasting of extreme streamflow is essential for effective reservoir management, including optimizing flood retention capacity and preparing for disaster response supplies. This study investigates whether the probabilistic forecasts of seasonal floods can be improved by integrating spectrally transformed hydroclimatic variables from the preceding season. Building on previous research, we proposed the spectral transformation technique to conditional covariates within a Generalized Extreme Value (GEV) modeling framework. Using streamflow observations from European catchments provided by the Global Runoff Data Centre (GRDC), we evaluated the role of transformed hydroclimatic covariates using Wavelet System Prediction (WASP) in enhancing seasonal flood forecasting skills. Results reveal that incorporating spectrally transformed covariates leads to improved forecasting skills measured by Ranked Probability Skill Score (RPSS) for a significant proportion of stations across Europe. Northern European catchments exhibit a stronger influence of climate covariates compared to Central and Western Europe. However, when transformed covariates are employed, teleconnections are enhanced across the continent, with notable improvements in the UK, Germany, and France. The hybrid WASP-GEV forecasting framework, integrating spectral transformation, significantly enhanced forecast skills with up to three months lead time. These findings underscore the importance of advanced data transformation and modeling techniques in improving the prediction of hydroclimatic extremes, offering practical implications for water resource management in a changing climate.
How to cite: Jiang, Z., Merz, B., and Sharma, A.: Enhancing Seasonal Flood Forecasts through Spectral Transformation of Hydroclimatic Covariates , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7620, https://doi.org/10.5194/egusphere-egu25-7620, 2025.
In Spain, there is an interest to incorporate seasonal forecasts into drought management, particularly by integrating the information these provide into the established system of drought indicators that are used to trigger drought management measures and to guide water infrastructure operations. So far, the use of such information to support operational decisions has been limited. Decision makers often quote poor forecast quality as well as information not being available at the scale commensurate to the drought indicators they use as a reason. Provision of forecast data that are credible, and at the spatial and temporal scales relevant to the decision-making process is key to improving uptake. Bias correction of seasonal forecasts plays a fundamental role in improving forecast quality. Several aspects inherent to the bias-correction processes may influence the degree of quality improvement from the perspective of user needs. An aspect that has, however, so far received little attention is the influence of the spatial resolution of both the forecast and the reference datasets that are applied on the quality of the decision-relevant indicators derived from the bias-corrected forecast.
In this study, we investigate the influence on forecast quality of the spatial resolution of both the forecast and the reference precipitation datasets in the bias correction process. We evaluate quality from the perspective of the indicators used in established operational drought management decisions. A Bayesian Joint Probability approach is used to bias-correct daily precipitation forecasts (ECMWF System 5) for a region within the Spanish Douro River Basin. We consider two resolutions of the forecast product (1° and 0.4°) and apply the bias correction using a historical dataset as a reference at three levels of catchment aggregation, namely, from the finest to the coarsest: independent sub-catchments used for the hydrological modelling of the region in the Douro Basin; aggregated sub-catchments used by the Douro Authority to determine the drought indicators used; and the entire catchment area of the main river in the region. Corrected seasonal precipitation forecasts at these three catchment aggregation scales are used to force a semi-distributed hydrological model to provide drought indicators derived from streamflow. Forecast quality is evaluated through a set of skill scores and user-centred metrics. Results show that the skill and reliability of bias-corrected precipitation forecasts do improve when compared to the raw forecast, with more substantial improvements at the coarser catchment aggregations and temporal resolutions. Results also show little difference in skill between the two resolutions of the forecast product. Furthermore, we show the influence of spatial resolution of both the forecast and the reference dataset on the skill improvement of the hydrological forecasts.
Results from this study contribute to the understanding and quantification of uncertainty in hydro-meteorological forecasts and in bias-correction and post-processing techniques. These findings are also useful in implementing seasonal forecasting and bias correction methods at scales appropriate to decision-making, thereby supporting operational drought management.
How to cite: Ramos Sánchez, C., Werner, M., De Stefano, L., and Schepen, A.: Influence of spatial resolution on forecast quality of bias-corrected seasonal hydro-meteorological forecasts from a drought management perspective, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-8075, https://doi.org/10.5194/egusphere-egu25-8075, 2025.
We propose here a data-driven approach to estimate prediction uncertainty for environmental models. The method is based on the analysis of prediction errors for past observations of the model output. It allows to estimate uncertainty for single model or multimodel predictions. The approach, called BLUECAT, operates by transforming a point prediction provided by deterministic models to a corresponding stochastic formulation, thereby allowing the estimation of a bias corrected expected value along with confidence limits. For multimodel predictions, at each prediction step we select the best performing model according to an uncertainty measure that is used as model selection criterion. We emphasise the value of BLUECAT for gaining an improved understanding of the underlying environmental systems and multimodel combination. Examples of applications are presented, highlighting the benefits attainable through uncertainty driven integration of several prediction models. A publicly available open software for the application of BLUECAT is available along with help facilities.
How to cite: Montanari, A. and Koutsoyannis, D.: Uncertainty estimation for environmental predictions: the BLUECAT approach and software, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5861, https://doi.org/10.5194/egusphere-egu25-5861, 2025.
In response to significant flood events, the Irish Government initiated the development of a national flood forecasting service in 2016. A key milestone in this initiative is the establishment of the Flood Forecasting Centre (FFC) within Met Éireann, which provides critical flood forecasting and advisory services to local authorities and emergency management stakeholders.
Met Éireann is currently advancing an operational fluvial flood forecasting system. The system integrates the HYdrological Predictions for the Environment (HYPE) model, developed by the Swedish Meteorological and Hydrological Institute, with the Delft-FEWS platform. Hosted on Microsoft Azure’s cloud computing service, this system utilizes real-time observational hydrometeorological data and ensemble Numerical Weather Prediction (NWP) forecasts from Met Éireann (Harmonie) and the European Centre for Medium-range Weather Forecasts (ECMWF). These ensemble forecasts enable the generation of probabilistic river discharge forecasts at multiple locations with lead times of up to seven days.
This poster highlights the core components of the fluvial forecasting system, detailing the development, calibration, and real-time operational workflow. Additionally, it explores system outputs, the role of ensemble forecasting, and future advancements, including the development of a robust verification system to evaluate forecast performance. This work represents a critical step forward in enhancing Ireland’s resilience to flood risks through robust and actionable forecasting capabilities.
How to cite: broderick, C., roberts, M., canavan, J., and lawlor, R.: Advancing Flood Forecasting in Ireland: Development and Implementation of an Operational Fluvial Forecasting System, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-3807, https://doi.org/10.5194/egusphere-egu25-3807, 2025.
Seasonal hydrological forecasts are of critical importance for the effective management of water resources, particularly in complex and vulnerable basins such as the Mediterranean area which have both a deficitary water regimen and a high anthropogenic pressure. Nevertheless, inaccuracies in meteorological inputs can propagate through hydrological models, amplifying uncertainties in flow predictions. It is imperative to rectify these forecasts, whether meteorological or hydrological, to enhance prediction reliability and provide robust data for informed decision-making. This study proposes an advanced framework for seasonal hydrological forecasting that integrates raw and corrected weather forecasts with distributed hydrological modeling and sophisticated post-processing techniques to enhance flow prediction accuracy in the Júcar River in Spain as a representative case study of the Mediterranean area.
The catchment hydrological model was implemented using the model TETIS v9.1 wich was calibrated and validated using observed records from 1981 to 2019 at seven control points. To do this, a split sample test was conducted using the period 2009–2019 for calibration, and the rest of the time series for validation (1981–2008). The process involved refining parameter maps to ensure good or acceptable performance at all control points. Meteorological data were sourced from the W5E5 dataset, downscaled to a 0.09° resolution using ERA5-Land. This improved the spatial and temporal resolution of the hydrological model. Once we have established an acceptable hydrological model,, the seasonal forecast hindcasts were evaluated using meteorological inputs from global forecasting systems, including ECMWF-SEAS5, CMCC_SPSv35, DWD_GCFS21, and MeteoFrance System8. To address uncertainties in the meteorological forecasts and their propagation to hydrological outputs, two complementary correction strategies were implemented. First, artificial intelligence(fuzzy logic) was applied to correct meteorological inputs before integration into the hydrological model, assuming errors originate solely from meteorological data and treating the hydrological model as a “perfect” simulator. Second, a hydrological error model was developed to identify and adjust discrepancies between simulated and observed flows, addressing systematic biases and errors in the hydrological simulation.
The results demonstrated that forecasts based on corrected meteorological inputs exhibited significant accuracy improvements compared to those using unprocessed inputs. The hydrological error model further enhanced prediction reliability by mitigating systematic biases. These findings underscore the effectiveness of combining meteorological forecasts with AI-driven corrections to address uncertainties, thereby improving the robustness of seasonal hydrological predictions. This study highlights the potential for integrating advanced correction techniques into seasonal hydrological forecasting frameworks, offering a replicable methodology for other basins with similar complexities. The proposed framework enhances both the reliability and applicability of forecasts, ensuring their relevance for effective decision-making in complex hydrological systems as the Mediterranean area. The improved accuracy of these forecasts provides a sound scientific support for adaptive water resource management, particularly in the face of increasing climatic variability and environmental changes.
Acknowledgments: This study was funded by the Colombian Ministry of Science, Technology, and Innovation (MINCIENCIAS) through the Call for Doctorates Abroad 885-2; by the Valencian Regional Government through the WATER4CAST 2.0 (CIPROM/2023/5) research project; and Spanish Ministry of Science and Innovation through the research project TETISPREDICT (PID2022-141631OB-I00).
How to cite: De León Pérez, D., Avila-Velazquez, D., Macian-Sorribes, H., Salazar-Galán, S., Pulido-Velazquez, M., and Francés García, F.: A Framework for Enhancing Seasonal Hydrological Forecasting in the Jucar River Basin (Spain), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12664, https://doi.org/10.5194/egusphere-egu25-12664, 2025.
The assessment of future discharge impacts on the Gandak River Basin is crucial for understanding potential climate change effects and planning effective water resource management. This study employs the Soil and Water Assessment Tool (SWAT) model integrated with machine learning techniques to evaluate and predict the future discharge patterns in the basin. The Gandak River Basin, a significant tributary of the Ganges, plays a vital role in regional agriculture, hydropower, and ecosystem services, making it imperative to understand the potential changes in its hydrological dynamics. The SWAT model, a comprehensive, semi-distributed hydrological model, simulates the effects of land management practices, climate variability, and water management strategies on water, sediment, and agricultural chemical yields in large complex watersheds. SWAT’s capability to incorporate various climatic inputs, land use, soil properties, and topography enables it to simulate hydrological processes with high accuracy. However, the complexity and non-linearity of hydrological processes often necessitate the incorporation of advanced data-driven techniques to enhance prediction accuracy and robustness. In this study, machine learning algorithms, including Random Forest, Support Vector Machines, and Neural Networks, are integrated with SWAT to improve the model’s predictive performance. These algorithms are trained on historical discharge data, climate variables, and SWAT-simulated outputs to capture the non-linear relationships and complex interactions within the hydrological system. The hybrid model leverages the strengths of both physically-based and data-driven approaches, providing a more comprehensive understanding of the future discharge scenarios under various climate change projections. The research involves hbias-correcting climate projections from General Circulation Models (GCMs) to derive high-resolution climate inputs for the SWAT model. Scenarios based on Shared Socio-Economic Pathways (SSPs) are employed to simulate future climatic conditions. The SWAT model is calibrated and validated using observed discharge data from the Gandak River Basin, ensuring the reliability of the simulations. Subsequently, the machine learning models are trained on the SWAT outputs and historical data, creating an ensemble approach to predict future discharge. Results indicate significant variability in future discharge patterns under different climate scenarios. The integrated SWAT and machine learning model captures the seasonal and inter-annual variability in discharge more accurately than the standalone SWAT model. The findings suggest potential increases in peak discharge events during the monsoon season, with implications for flood risk management. Conversely, reduced discharge during the dry season could impact water availability for agriculture and domestic use, necessitating adaptive water management strategies. The study highlights the importance of combining physically-based hydrological models with machine learning techniques to enhance the prediction of hydrological responses to climate change. The integrated approach provides valuable insights for policymakers and stakeholders in the Gandak River Basin, aiding in the development of sustainable water resource management plans to mitigate the adverse impacts of future climate variability. This research underscores the need for continuous monitoring, adaptive management, and the incorporation of advanced modeling techniques to address the complexities of climate change impacts on river basins.
How to cite: Jha, A., Bansal, J. K., and Gupta, N. C.: Assessing the impact on Future Discharge on Gandak River Basin Using SWAT Model and Machine Learning Techniques, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-835, https://doi.org/10.5194/egusphere-egu25-835, 2025.
Floods are among Earth's most widespread, frequent, and destructive natural hazards. In highly responsive
catchments, daily streamflow predictions will underestimate flood hazards. For example, peak flows occurring
on sub-daily timescales caused hundreds of fatalities and billions of Euros in damages in the devastating floods
in Germany in 2021 and Spain in 2024. Particularly in small and mesoscale catchments: 1) peak flows may
last only a few hours, so forecasts of daily flows can greatly underestimate flood peaks ; 2) A landscape's
responsiveness to precipitation depends critically on how wet it is; thus, it is essential to accurately model the
wetting and drying of the catchment, and hourly streamflow is needed to capture and understand the
hydrological processes in the rising limb of the hydrograph; and 3) the dominant processes affecting shortterm
predictions are not necessarily the same as those affecting streamflow at longer time scales. For example,
over longer time scales, predictions become more a question of mass balance, rather than dynamics and routing,
while the opposite is true for short-term predictions.
Thus, reliably assessing flood hazards requires understanding hydrologic responses at hourly time scales. But
paradoxically, hourly predictions have received relatively less focus. In this work we use a conceptual
hydrological model to obtain deterministic hourly predictions and estimate its uncertainty using a residual error
model. Case study catchments include hydrologically diverse catchments in Europe and the USA. We consider
bias, heteroscedasticity and autocorrelation by employing the Box-Cox transformation, autoregressive (AR)
and moving average models (ARMA) models. The log transformation was in general the most recommended
option, in combination with an AR3 model.
This work advances streamflow prediction by developing statistically rigorous methods for postprocessing the
residuals of conceptual models at the hourly time scale.
How to cite: Prieto, C., Kavetski, D., Fenicia, F., Kirchner, J., and Álvarez, C.: Towards achieving reliable probabilistic hydrological predictions at the hourly scale, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2946, https://doi.org/10.5194/egusphere-egu25-2946, 2025.
Advanced Hydrological Ensemble Prediction Systems (HEPSs) offer significant potential to enhance real-time water management by providing probabilistic ensemble water forecasts that can help dam operators better anticipate and mitigate risks. However, fully utilizing HEPS forecasts in real-time decision-making presents major challenges. To address this for dam operations, we developed an integrated HEPS-optimization program to determine the required dam releases to meet a downstream target flow, considering short-term ensemble tributary inflow forecasts. We specially designed the program to have the ability to explicitly limit downstream flood risk through chance constraints. This ability is highly desirable for more effective risk-based operations but is lacking in the large majority of existing methods. A complicating factor however was that the ensemble nature of the tributary inflow forecasts necessitated formulating the chance constraints as discontinuous mixed-integer equations, which makes the problem nondeterministic polynomial-time hard. Thus, to solve the program, we adopted an innovative approach combining a novel ranking mechanism with nonlinear programming. We favoured this approach over conventional branch-and-bound methods and stochastic dynamic programming as it is considerably faster. We evaluated the viability of our methods using a case study of Hume Dam and Lake Mulwala in the Murray-Darling Basin, Australia. The results demonstrate their efficacy.
How to cite: Ng, T. L. S., Robertson, D., and Bennett, J.: Integrating Hydrological Ensemble Prediction System and Optimization for Limiting Downstream Flood Risk in Dam Operations, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-5312, https://doi.org/10.5194/egusphere-egu25-5312, 2025.
Ensemble forecasting is a powerful tool for supporting informed decision-making in managing multi-hazard risks associated with tropical cyclones (TCs). While TC ensemble forecasts are widely utilized in operational numerical weather prediction systems, their potential for disaster prediction remains underutilized. Here we propose a novel, efficient, and practical method to extract meaningful multi-hazard worst case scenarios (MHWCS) from a large ensemble TC forecast of 1000-members for the first time. We perform the simulation of TC Hagibis (2019) using the Japan Meteorological Agency's (JMA) nonhydrostatic model. The simulated atmospheric predictions were serving as inputs for JMA’s operational flood forecast model, as well as statistical storm surge and gust wind models. These models estimate river flooding, storm surge, and wind hazard intensities in Tokyo. By accounting for uncertainties in ensemble multi-hazard forecasts, we objectively demonstrate that Pareto-optimal solutions can effectively identify the meaningful MHWCS. These solutions illustrate complex trade-offs among competing hazard components across various forecast locations. Notably, the meaningful MHWCS do not necessarily represent the most extreme values for individual hazards but instead maximize hazard intensities relative to the ensemble mean, collectively leading to significant disaster impacts. Our findings further underscore the importance of evaluating Pareto-optimal solutions to assist risk managers in understanding how combinations of TC meteorological variables—such as track, translation speed, intensity, size, and rainfall—shape worst-case scenarios. For instance, meaningful MHWCS forecasts tend to exhibit moderate meteorological characteristics comparable to the ensemble mean, with variability in translation speed emerging as the strongest single predictor of single-hazard worst-case scenarios.
How to cite: Islam, Md. R., Oizumi, T., Duc, L., Ota, T., Kawabata, T., and Sawada, Y.: Advancing Multi-Hazard Analysis: Worst Case Scenarios from Ensemble Tropical Cyclone Forecasts, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14562, https://doi.org/10.5194/egusphere-egu25-14562, 2025.
In France, groundwater is one of the most important resources for industry, agriculture, and drinking water. While severe droughts affecting groundwater are becoming more frequent, the development of forecast has become essential for stakeholders. The hydro-meteorological platform Aqui-FR (Vergnes et al., 2020), which integrates different regional groundwater models, is coupled with atmospheric reanalysis and downscaled seasonal forecasts (Willemet et al., 2022) to achieve this goal.
However, biases and errors in the models used still affect the predicted initial conditions (IC), limiting the potential for operational use. To overcome these problems, a data assimilation (DA) scheme has been developed within the Aqui-FR workflow. The analysis step focuses on state estimation, and more specifically on piezometric (groundwater) levels during a reanalysis run. An Ensemble Kalman Filter (EnKF ; Evensen, 1994) has been implemented in a Python library (aquida) to set up a sequential DA. Two inflation methods and two localisation methods (quasi-Gaussian distance-based and spatial autocorrelation-based) are used.
To analyse the efficiency of the DA scheme, this first study focuses on one of the regional models of the Aqui-FR platform, the Somme basin model which uses the MARTHE hydrogeological computer code (Thiéry et al., 2020) to simulate both piezometric levels and river discharge. In situ piezometric data from monitored wells will be assimilated.
Preliminary results obtained from our numerical experiments show the benefit of DA on groundwater state estimation with a regional model (mean RMSE reduced from 4.26 to 0.32), even with spatially sparse data. When assimilation is stopped, the analysis shows an impact on state estimation up to a seasonal time step (mean RMSE about 2.9 after 180 days without assimilation), which is encouraging for forecast improvements. Nevertheless, in regions of the model domain where the initial calibration is too poor, the correction shows less persistence and the dynamics of the model appear to be driven by parameters rather than forcing. To improve the piezometric estimation in these areas, we plan to implement a two-step DA with parameter estimation prior to state estimation.
References
Evensen, G. (1994). Sequential data assimilation with a nonlinear quasi-geostrophic model using Monte Carlo methods to forecast error statistics. Journal of Geophysical Research: Oceans, 99 (C5), 10143–10162. https://doi.org/10.1029/94JC00572
Thiéry, D., Picot-Colbeaux, G., & Guillemoto, Q. (2020). Guidelines for MARTHE v7.8 computer code for hydro-systems modelling (English version) (tech. rep. No. BRGM/RP69660-FR). BRGM.
Vergnes, et al. (2020). The AquiFR hydrometeorological modelling platform as a tool for improving groundwater resource monitoring over France: Evaluation over a 60-year period. Hydrology and Earth System Sciences, 24 (2), 633–654. https://doi.org/10.5194/hess-24-633-2020
Willemet, J.-M., et al. (2022) Aqui-FR: Towards a hydro-geological seasonal forecasting system for metropolitan France. In: IAHS-AISH Scientific Assembly. IAHS2022-525. Montpellier, France: Copernicus Meetings. https://doi.org/10.5194/iahs2022-525
How to cite: Manlay, A., Vergnes, J.-P., and Habets, F.: On the need for better groundwater initial conditions estimation in seasonal forecasts: a data assimilation scheme for Aqui-FR hydrometeorological modelling platform. Example with the regional case study of the Somme basin (France)., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-16561, https://doi.org/10.5194/egusphere-egu25-16561, 2025.
Accurate reservoir level prediction is vital for effective reservoir operation. While an overprediction of the reservoir levels results in the release of excess water than required, an underestimation leads to insufficient water supply. This affects the multiple purposes served by the reservoir, such as domestic and municipal water supply, irrigation, hydropower generation, and flood control. However, predicting reservoir levels accurately is complex and challenging owing to the errors arising from the hydrological and routing models. This affects the accuracy of the predicted reservoir levels and incorporates uncertainty. Thus, it is vital to explore measures to reduce the error in the predicted reservoir levels to improve their reliability. The study thus proposes a novel error correction modelling framework for reducing the prediction uncertainty in the reservoir levels. For this endeavor, the state of the art of machine learning models is exploited. The proposed framework integrates an optimization technique with machine learning models to reduce the error in the predicted reservoir levels. The framework was tested on the Pong reservoir, India, and evaluated using several performance indices including the normalized root mean square error (NRMSE), Nash Sutcliffe efficiency (NSE), and percentage of coverage (POC). The evaluation revealed improvements in accuracy and a reduction in uncertainty of predicted reservoir levels. For example, the NRMSE of the predicted reservoir levels improved from 0.132% to 0.002% and 0.416% to 0.397% during calibration and validation respectively, while the percentage of coverage improved from 45% to 77.5% (calibration) and from 27.27% to 36.36% (validation). The framework thus has the potential to improve reservoir operational control and associated decision-making.
How to cite: Roy, A., Patidar, S., J. Adeloye, A., and Kasiviswanathan, K. S.: A novel error correction modelling framework for reducing prediction uncertainty in reservoir levels for operational control, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-639, https://doi.org/10.5194/egusphere-egu25-639, 2025.
Droughts have multiple definitions for a range of drought types, and related indicators, in science alone. For persons experiencing and having to cope with droughts, individual, sectoral, and local definitions vary even further. Hydrometeorological sub-seasonal to seasonal forecasts of droughts may help to better plan, communicate, and implement mitigation measures. Forecast skill of such forecasts for droughts may depend on the drought definition or definitions analysed.
This research, therefore, analyses in depth, drought impacts, mitigation measures, decision making processes and the potential added value of using forecasts, for a case study in the Netherlands: Rijnland. This is a low-lying flat area in the West, mostly below sea-level, with a dense irrigation and drainage water system to maintain surface water levels in a narrow target range along with its water quality. Both hydrological droughts in the Rhine river basin, and meteorological droughts locally in Rijnland, affect and may trigger drought mitigation actions in Rijnland. Case study drought definitions for early warning are expressed in terms of time-varying lower thresholds for Rhine discharge, and thresholds of potential precipitation deficit varying for different levels of alert.
Forecast skill and potential added value for case study specific mitigation actions of sub-seasonal to seasonal hydrometeorological reforecasts, both directly available and AI-enhanced, are presented and intercompared with the aim to arrive at well-informed recommendations for their use or non-use in the case study of Rijnland.
How to cite: van Andel, S. J. and Bertini, C.: Forecast skill and potential added value for drought mitigation actions of sub-seasonal to seasonal AI-enhanced hydrometeorological forecasts: case study of Rijnland, the Netherlands, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19967, https://doi.org/10.5194/egusphere-egu25-19967, 2025.
Posters virtual: Tue, 29 Apr, 14:00–15:45 | vPoster spot A
EGU25-13991 | Posters virtual | VPS9
Ensemble Approach for Hydrological Forecasting Based on Recurrent Neural Networks and Complex NetworksTue, 29 Apr, 14:00–15:45 (CEST) | vPA.15
The São Francisco River Basin is crucial for Brazil’s agriculture, hydropower, and water security. However, climate change has intensified challenges like reduced water flow and frequent extreme events, threatening its socio-economic sustainability. This study aims to forecast flow in the São Francisco River Basin, enabling proactive decision-making to mitigate risks associated with both droughts and floods. To address these challenges, this study propose a novel methodology based on Artificial Intelligence (AI), combining Recurrent Neural Networks (RNN) and complex network techniques. The method creates new features and assigns importance weights to enhance the algorithm’s ability to generate probabilistic flow forecast. The results are promising, demonstrating the method’s ability to deliver accurate probabilistic forecasts. This research can support risk mitigation strategies and improve water resource management in the São Francisco Basin. Additionally, the proposed framework is scalable, offering potential applications to other critical watersheds facing similar challenges
How to cite: Caseri, A., Aparecido Rodrigues, F., and Victal Cerqueira, M.: Ensemble Approach for Hydrological Forecasting Based on Recurrent Neural Networks and Complex Networks, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13991, https://doi.org/10.5194/egusphere-egu25-13991, 2025.
