In regions such as South Korea, where rainfall is irregular over time, it is important to operate dams and reservoirs to regulate the supply and demand of water, ensure proper irrigation water discharge to downstream agricultural areas, and prevent flood damage. However, during periods of heavy rainfall, it becomes difficult to operate and manage reservoirs normally, and time-series AI models can be used to produce reservoir storage rate forecasts to help smoothly operate reservoirs and manage water balance. In this study, multivariate time series forecasting was performed for the top 46 reservoirs with effective storage in Korea, using historical storage, precipitation, Julian day, and estimated inflow and outflow calculated using an expert knowledge-based rainfall-runoff model as input variables to predict storage from 1 day up to 20 days-ahead, and quantitative evaluation of reservoir storage rate predictions was performed using MAE (mean absolute error), MBE (mean bias error), RMSE (root mean square error), and CC (correlation coefficient) metrics. Predicting reservoir storage rate by considering only seasonal and meteorological effects reduces accuracy during the rainy season. However, when the estimated inflow and outflow of reservoirs, derived using an expert knowledge-based rainfall-runoff model, were additionally incorporated as inputs into the time series model, the average MAE of the 46 reservoir storage forecasts improved from 0.088%p to 0.240%p, particularly enhancing the low forecast performance in the rainy season. Furthermore, an ensemble of time series models with recurrent neural network structure, which has strengths in short-term forecasting and transformer structure, which has strengths in medium-term forecasting produced better reservoir storage rate predictions than the single model in both short-term and medium-term. The MAE and RMSE averages of the ensemble model's 1-day-ahead reservoir storage rate predictions for 46 reservoirs were 1.384% and 2.496% respectively, an improvement of 0.573% and 0.644% over the single model, and the statistical superiority of the ensemble model increased as the number of days in the future was predicted. The importance of the input variables of the ensemble model was evaluated, and the historical reservoir storage rate was the most important with 67.24%, followed by the estimated inflow and estimated outflow with 12.36% and 8.73%, and the sum of the importance of the two variables calculated through rainfall-runoff modeling was 21.09%. By using variables that reflect the management practices of the reservoir, the model provided information that the model could not learn from the reservoir storage rate, meteorological data, and seasonal data alone. This study enables stable reservoir operation throughout the year, even in areas with irregular rainfall, and is expected to improve agricultural stability in downstream irrigated areas and prevent rainfall flooding.
This work is supported by the Korea Agency for Infrastructure Technology Advancement (KAIA) grant funded by the Ministry of Land, Infrastructure and Transport (Grant RS-2022-00155763).
How to cite: Park, J. and Lee, Y.: An ensemble of AI time-series models for reservoir storage rate in South Korea: Accuracy improvement using expert knowledge-based rainfall-runoff modeling, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7992, https://doi.org/10.5194/egusphere-egu25-7992, 2025.
Reservoirs are a vital aspect of water resource management in many regions worldwide, used to meet domestic and irrigation demand, for hydropower, flood control, and maintaining river flow for ecological and navigational purposes. It is important to have a reliable forecast for reservoir status to ensure efficient operation of individual reservoirs and the wider water resource system, particularly during unusually wet or dry periods. This forecasting will become increasingly important under a changing climate and growing demand for water and hydropower. In this work, we focus on sub-seasonal to seasonal forecasts, providing vital information for water users and enabling them to make informed decisions on management strategies with seasonal lead-times.
There has been research on forecasting reservoir status (i.e. reservoirs storage, inflow, outflow, level, or storage anomaly) at different lead-times with machine learning (ML) methods, with most studies producing reservoir-specific models. In this work, we produce an Extra Trees (ET) regressor model for multiple reservoirs over Europe, trained on historical monthly storage, rainfall and temperature, along with static catchment characteristics from the relevant CAMELS (Catchment Attributes and Meteorology for Large-sample Studies) datasets. This model is used to forecast storage at one- and three- month lead times for reservoirs in the region, including reservoirs unseen by the model in the training period. A multi-reservoir model makes use of all the available data, allowing forecasts for reservoirs with limited historical data, and improves prediction performance under extremes compared to a single-reservoir model.
The model has already shown promising results for predicting reservoir storages using observed climatic inputs, and this work will investigate the model skill in forecast mode, using ensemble climate hindcasts, for reservoirs across Europe. It is anticipated that this model will provide a computationally- efficient forecasting tool with relatively low input data requirements that can be used to forecast reservoir storage in the modelled area, with potential to expand to a global extent.
How to cite: Baron, H., Chengot, R., Rickards, N., and Keller, V.: A multi-reservoir machine learning model for forecasting reservoir storage at monthly and seasonal timescales, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9347, https://doi.org/10.5194/egusphere-egu25-9347, 2025.
In the Euro-Mediterranean region, summer droughts present significant challenges for various socio-economic sectors, raising the need for reliable seasonal drought forecasts to support proactive water resource management. This study evaluates the skill of the latest seasonal forecast systems from the Copernicus Climate Change Service in predicting summer droughts, using the Standardized Precipitation Evapotranspiration Index (SPEI) to characterize drought events. Using a systematic multi-metric evaluation framework that includes both deterministic and probabilistic scores, we benchmark individual systems and their multi-model ensemble (MME) to identify patterns of predictive skill across regions and lead times. The findings reveal that when SPEI forecasts are initiated at the onset of the summer season, all models exhibit on average positive correlations with observed dry conditions, reflecting also good quality in terms of accuracy, reliability, and discrimination skills, though with local variability. The added value of dynamical models compared to climatology-based heuristic prediction methods declines significantly for forecasts initialized one month earlier. At all lead times performance is better for all models in Southern Mediterranean areas, indicating higher predictability of SPEI in that region compared to Northern Europe. By highlighting the grid points where SPEI seasonal forecasts hold significant predictive value, this study provides actionable insights for leveraging these products in decision-making processes. When a non-locally tailored analysis is needed, the MME offers the most robust drought forecasting solution, always demonstrating more widespread significant skill with respect to single models up to a 1-month lead time, covering much of the Mediterranean region. Beyond this horizon, significant skill becomes limited, resulting in forecasts that are neutral compared to the heuristic approaches. Notably, individual models demonstrate localized lead time-dependent strengths that may make them preferable to the MME in specific cases where tailored predictions are required.
How to cite: Cerato, G., Bellomo, K., and von Hardenberg, J.: Summer drought predictability in the Euro-Mediterranean region in seasonal forecasts, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13167, https://doi.org/10.5194/egusphere-egu25-13167, 2025.
Droughts, prolonged heatwaves, heavy rainfall and multiple large-scale floods - recent years have shown that global climate change requires more sustainable and timely water management at all levels. In particular, for the optimized use of water resources for irrigation or hydropower generation, it is essential to know their likely availability in the coming months anywhere in the world. This sub-seasonal to seasonal time range is covered by seasonal forecasting systems such as SEAS5 developed by the European Center for Medium-Range Weather Forecasts (ECMWF). These systems have the potential to provide important data for improving water management practices. However, without a bias correction, the data deviate strongly from the actual data. We have shown for several regions of the world that the Bias Correction and Spatial Disaggregation (BCSD) method can significantly improve predictive capability. By storing fixed cumulative distribution functions (CDFs) and parallelization, we were able to extend the system from the regional to the global level, i.e. to produce BCSD predictions for the entire globe and present this version at EGU24.
Our next step is to evaluate the resulting seasonal forecasts in terms of their predictive quality. This evaluation is carried out using several measures of quality, including the Continuous Ranked Probability Skill Score (CRPSS) and the Brier Skill Score (BSS). To illustrate this, the strengths and weaknesses of the bias-corrected seasonal forecasts are highlighted using two regions. The focus is on the Sahel region, which has a lower forecast quality despite its high social relevance, and on the Lake Victoria catchment area, for which a high forecast quality is achieved. The aim is to achieve as precise an assessment as possible of the global forecast quality, which allows a realistic assessment of the forecasts, particularly in regions with strong fluctuations in water availability. By providing this bias-corrected forecast data in near real time together with an analysis of its quality, better estimates will be available for direct use by water managers or as input for subsequent modeling processes.
How to cite: Weber, J. N., Lorenz, C., Schober, T., Wiegels, R., Chwala, C., Fersch, B., and Kunstmann, H.: Analysis of global bias-corrected seasonal Forecasts: Where do the strengths and challenges lie?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2867, https://doi.org/10.5194/egusphere-egu25-2867, 2025.
The US Secure Water Act of 2010 requires several US agencies to report to Congress every five years on future water-related mission vulnerabilities. Over the last 15 years, 21st century climate projection datasets from the Coupled Model Intercomparison Projects (CMIP) have been downscaled and used to drive hydrologic and streamflow scenarios across the Contiguous United States (CONUS). The resulting datasets form input for federal and state agency planning, guidance and policy, for water resources applications from watershed to regional scales, and for the climate-water research community. The advent of CMIP6 has triggered the co-development of new, updated hydrologic modeling for future hydroclimate impact projections, which is proceeding via a multi-agency effort that integrates researchers with stakeholders from US federal water, climate and energy agencies. The effort has lately spurred interest in a related trans-boundary joint hydroclimate science effort between the US and Canada. This effort uses the process-oriented SUMMA land/hydrology model and mizuRoute channel routing model, which have been configured for CONUS and adjoining watersheds at a USGS HUC12 (and MERIT-Hydro) watershed resolution, a contrast to earlier grid-based modeling approaches. Several hundred CMIP6 future climate scenarios are being downscaled to drive future hydrologic assessments that are tailored to water agency planning needs.
This work necessitated the creation of new strategies to upgrade existing capabilities in continental-scale process-based hydrological modelling and projections, which have been undermined by poor calibration in prior iterations. Notable innovations included a powerful new large-sample parameter estimation approach based on machine-learning (ML) emulators; creating extended (CAMELS-like) large-sample catchment datasets for model calibration and validation (using both natural and reconstructed historical streamflow observations); creating a new CONUS-wide multi-decadal high-resolution surface meteorological (forcing) dataset, derived using ML methods; and the use of water management guided performance metrics to inform model training and evaluation. This presentation summarizes the new CMIP6 hydroclimate dataset initiative, and highlights the critical role of integrated researcher-stakeholder engagement in achieving fit-for-purpose and actionable large-domain hydrology outcomes.
How to cite: Wood, A., Tang, G., Farahani, M., Mizukami, N., Mueller, C., Frans, C., McGuire, M., and Thames, B.: Co-developing CONUS-wide current and future hydroclimate projections to support US agency water security initiatives, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14598, https://doi.org/10.5194/egusphere-egu25-14598, 2025.
Robust adaptation to hydrological climate change requires an assessment of possible climate and hydrological futures at different scales. Explore2 is a recent multimodel ensemble of hydrological projections developed to accompany local adaptation plans for metropolitan France and Corsica, a 550,000 km2 wide territory with a large diversity of climate and hydrological regimes (Sauquet et al. 2024). The Explore2 ensemble provides transient projections of daily river flow for more than 4’000 locations on French rivers. Projections have been obtained for three RCP emission scenarios with 4 to 9 hydrological models (HMs) driven by 36 bias adjusted regional climate projections (36 EUROCORDEX projections obtained for a number of different GCM/RCM combinations, Marson et al. 2024).
Explore2 projections, making no exception, come with sometimes large uncertainty (Evin et al., 2024). This uncertainty has been characterized and analyzed with Qualypso (Evin et al. 2021) for different climate and hydrological metrics. Qualypso is an advanced ANOVA approach, based on the quasi-ergodic assumption for transient climate projections (Hingray et Said, 2014) and applicable for unbalanced datasets. It allows to disentangle and prioritize the different components of uncertainty, namely emission scenario uncertainty, the different components of model uncertainty (GCM uncertainty, RCM uncertainty, HM uncertainty) and uncertainty due to climate internal variability.
In this work, we examine the following questions:
- What are the projected changes for different climate and hydrological metrics for metropolitan France and Corsica (e.g. precipitation, annual discharge, high and low flows) and how strong do the modelling chains agree on projections?
- How do scenario uncertainty and the different components of model uncertainty contribute to the total uncertainty in projections and what additional variability is introduced by internal variability of climate?
- What are the main effects of each model compared to the others and are there highly contrasting chains?
- How do the results depend on location and/or hydroclimatic context?
- What messages can be conveyed to stakeholders about the future of climate and hydrology in France?
References:
Evin et al. 2021. Earth System Dynamics. https://esd.copernicus.org/articles/12/1543/2021/
Evin et al. 2024. Recherche Data Gouv. https://hal.science/hal-04609542
Hingray and Saïd. 2014. J.Climate. https://doi.org/10.1175/JCLI-D-13-00629.1
Marson et al. 2024. Recherche Data Gouv. https://hal.science/hal-04443633v1
Sauquet et al. 2024. Recherche Data Gouv. https://doi.org/10.57745/J3XIPW
How to cite: Hingray, B., Evin, G., Sauquet, E., Reverdy, A., and Ducharne, A. and the Explore2-HyMet: Uncertainty sources in a large ensemble of hydrological projections across diverse hydroclimates: characterization and take-home messages for stakeholders, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18641, https://doi.org/10.5194/egusphere-egu25-18641, 2025.
The lack of saliency, credibility, and legitimacy of information in climate services constrains their use in decision-making. Co-creation processes involve end users, purveyors, and providers in an iterative approach to develop tailored climate services that are useful and usable. Such approaches facilitate contextual understanding, integrate different knowledges, and are key in bridging the gap between innovation and use.
Co-creation approaches have been applied in climate service development over the past years. However, there has been little exploration on whether and how the process can be organised to ensure that the co-created climate services are sustainably used. Unpacking this relationship between co-creation and use enables an understanding of the factors to consider, and pathways to follow when co-creating climate services to effectively upscale innovation and sustainable use in decision-making.
This research aims to identify the pathways through which co-creation processes contribute towards the use of climate services in decision-making. We follow and critically evaluate an ongoing co-creation process for the development of a Drought Early Action Protocol (and the revisions thereof) in Lesotho. We apply the Contribution Analysis method to evaluate the process with key stakeholders from the Lesotho Meteorological Services, the Disaster Management Agency, and the Lesotho Red Cross Society. Preliminary findings suggest that a combination of i) embedding the climate service in an already existing decision-making framework, ii) actively building the capacity of institutional staff to use the climate service, and iii) providing avenues for maintenance after the co-creation process, is a likely pathway a co-creation process may take towards ensuring use of the climate service. This work is beneficial to practitioners and researchers as it provides an empirically grounded explanatory account of the relationship between co-creation and the use of climate services.
How to cite: Nyamakura, B., Werner, M., Masih, I., Castellana, D., and van den Homberg, M.: Enhancing the usefulness and usability of Drought Early Action Protocols through co-creation , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-17686, https://doi.org/10.5194/egusphere-egu25-17686, 2025.
The increasing intensity and duration of droughts threatens water availability with greater frequency (Ipcc 2022), requiring the adoption of measures in response to these extreme weather events (Berrang-Ford et al. 2021). The lack of hydroclimatic information adapted to users’ needs constrains the adaptive capacity of decision-makers (Lee et al., 2022). To bridge this gap, it is necessary to combine available scientific-technical information with local knowledge through co-production processes that adapt the information to the local context and to the user’s needs (Norström et al. 2020).
In the I-CISK project (https://icisk.eu/), we developed a pre-operational user-centered climate service with hydrogeological information in the Andalucia-Los Pedroches Living Lab, located in northern Cordoba, Spain. The aim of this work is to develop a hydrogeological model of Los Pedroches hard rock aquifer system and its relationship with evolving climate. The availability of hydrological and hydrogeological data in the region is limited, both in terms of the number of measuring points and of the length of available data time-series. The model has been built by integrating scientific and local information using MODFLOW-NWT (USGS, 2022). In this work, local knowledge is understood as personal experience and local ecological knowledge regarding hydrological and hydrogeological information (van den Homberg et al., 2023). First, we conducted an exhaustive review of the information available in the literature, official databases and previous technical studies. Second, we designed a survey to collect local knowledge regarding groundwater functioning and characteristics through semi-structured interviews with users, focus groups and workshops. Third, to complement the limited availability of groundwater level data, we conducted four field campaigns to build a water table database for model validation.
Technical and scientific research enabled the generation of a structured set of data and the identification of knowledge gaps. These gaps were filled based on local knowledge gathered and verified with available hydrogeological information to define the model's physical environment. Local knowledge enabled the adjustment of model boundary conditions, the estimation of the thickness of the aquifer layer, the aquifer-river network connections, and the influence of dikes and fractures on subsurface flow. As in other contexts (Guodaarmarti et al., 2021; Habté et al., 2021), local knowledge integration is key to improving the design of climate change adaptation measures. The integration of these different sources of data and qualitative information into a climate service is expected to provide a more comprehensive characterization of the aquifer and its functioning to improve decision-making processes regarding water resources use.
How to cite: Ropero, N., De Stefano, L., and Heránde-Mora, N.: Local knowledge integration to develop user tailored hydroclimatic service, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19791, https://doi.org/10.5194/egusphere-egu25-19791, 2025.
The use of weather and climate information, or data and insights relating to both short- and long-term weather patterns in a specific region, has been encouraged to better understand, address, and mitigate the impacts and challenges presented by climate change. However, despite ongoing efforts to improve the development and availability of climate information, it is not well understood whether this information is made obvious to relevant users, and the extent to which climate information is utilised for improved decision-making. Further, addressing the knowledge gap of why the uptake and use of climate information is low or not done despite being made available to users would be a valuable new contribution and a space for behavioural science, as it would question the notion that the provision of knowledge automatically leads to action.
This research shares insights from the Behavioural Adaptation for Water Security and Inclusion (BASIN) project, which is funded by UK aid from the UK government and by the International Development Research Centre (IDRC), Canada, as part of the Climate Adaptation and Resilience (CLARE) research programme. This project underscores how more inclusive water security and equitable adaptation can be supported, and one of its core research questions focuses on examining community perceptions of climate information, whether and how actions are taken as a response of available information, and reasons why climate information was not used. This presentation summarises these findings based on responses to a series of focus group discussions and key informant interviews undertaken in Tanzania, Burkina Faso and Malawi. For instance, it underscores barrier and enabling factors affecting climate information use and highlights how climate information could be better packaged for increased use in crop planning and enhanced agricultural production as well as flood and drought management. Such insights thus offer a context for the need for behavioural interventions that could be helpful to assist improved decision-making and community practices on water security and climate adaptation.
How to cite: Dookie, D. S., Barry, D., Damiba, L., Gungulundi, V. T., Lalika, C., Komakech, H., Monjerezi, M., and Vincent, K.: Behavioural insights on climate information uptake in Tanzania, Burkina Faso and Malawi, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13941, https://doi.org/10.5194/egusphere-egu25-13941, 2025.
Decision Support Systems (DSS) play a critical role in climate adaptation by providing tools to guide decision-making processes. However, most existing DSS primarily focus on delivering scientific information, often reflecting a scientism-driven and enlightenment-oriented perspective that assumes knowledge dissemination alone will lead to optimal decisions. This approach must frequently pay more attention to decision-makers' diversity, contexts, and the participatory mechanisms necessary to incorporate local knowledge and values. As a result, DSS in climate adaptation often fails to bridge the gap between scientific expertise and the lived realities of citizens and stakeholders.
This study examines citizen and stakeholder knowledge integration within DSS in climate adaptation platforms through a comprehensive evaluation framework. The research begins by questioning what constitutes a “decision” in climate adaptation, who the decision-makers are, and how decisions are made. By addressing these foundational questions, the study highlights the limitations of current DSS, which are primarily information-centric, and explores their implications for participatory governance.
This research evaluates the functionality of over 50 global and domestic climate adaptation platforms based on interactivity, accessibility, contextual relevance, and capacity for fostering collaboration through systematic classification and mapping. It also emphasizes the role of participatory tools, citizen science, and co-creation practices in transforming DSS from mere data providers into platforms for collaborative decision-making.
How to cite: Song, K. and Yang, S.-B.: Integrating Citizen and Stakeholder: Mapping Platforms and Decision Support Systems for Climate Adaptation, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2426, https://doi.org/10.5194/egusphere-egu25-2426, 2025.
Posters on site: Wed, 30 Apr, 14:00–15:45 | Hall A
Spatial rainfall variability directly impacts hydrological basin responses, especially in regions with complex interactions between hydrometeorological phenomena and physical factors. Understanding this influence supports the development of more accurate flow forecasting models, enhancing practical applications in water resources management. However, current studies often overlook the effect of rainfall spatial distribution on forecasting outcomes, despite its critical role in shaping hydrological responses. Such forecasts are essential for planning water distribution across sectors like human consumption, agriculture, and energy generation. In Latin America and the Caribbean, where hydropower supplies approximately half of the region's electricity, accurate forecasts are crucial. Ecuador, for instance, relies on hydropower for over 85% of its energy needs, underscoring the necessity for reliable hydrological forecasts, particularly in the Andean tropical mountain basins where major hydropower plants are located. One of the most important hydropower systems in the country, supplied by its largest reservoir, currently lacks an operational hydrological forecast to support its management, which is urgently needed due to the ongoing drought conditions impacting hydropower production. To address this, we developed data-driven models, known for their ability to handle data complexity and outperform conceptual or physical models in scenarios with high complexity and limited precise data, which is characteristic of our study area. These models analyze the influence of spatial rainfall variability on reservoir outflow forecasting using an interpretable approach to identify the most impactful rainfall data configurations. We employed satellite-based rainfall data from IMERG and GSMaP, which have shown promising results in nearby basins, across five top-down configurations: mean rainfall, climatological rainfall regions, seasonal clusters, travel time regions, and spatially distributed data. The models, based on neural networks including (RNN) Recurrent Neural Networks and Long-Short Term Memory (LSTM) architectures, are configured to provide forecasts from hourly to daily scales across these scenarios, supporting practical operational applications. Initial experiments indicate strong model performance, with NSE values ranging from 0.9 to 0.45 for hourly forecasts at lead times from 3 to 24 hours. To enhance interpretability, methods such as SHAP (SHapley Additive exPlanations) are applied to understand how rainfall data conditions model performance under different hydrological scenarios. With this approach, we aim to identify the optimal rainfall data setups for improved forecasting models in these basin settings.
How to cite: Merizalde, M. J., Corzo, G., Samaniego, E., Muñoz, P., and Célleri, R.: Analyzing the influence of spatial rainfall variability on reservoir outflow through interpretable data-driven modelling: an application in Tropical mountain basin, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-942, https://doi.org/10.5194/egusphere-egu25-942, 2025.
Hydrological forecasting is an essential tool for water resource management, enabling predictions of the future state of water resources in a catchment. Since some years, the forecast horizon / lead time is increasing. The demand for reliable seasonal hydrologic forecasts is significant and various applications for water resources management are increasingly important. Integrating seasonal forecast results into decision-making processes have become vital for both short- and long-term water management across various sectors, including energy, water supply, agriculture, urban planning, infrastructure, and disaster preparedness.
This paper assesses seasonal streamflow and sediment forecasting as a critical component of effective water resource management. An effective seasonal water resources forecasting system requires an evaluation of both numerical weather prediction (NWP) models and hydrological models to accurately represent atmospheric and hydrological conditions in a specific region. This study evaluates the ECMWF-SEAS5 precipitation product in conjunction with a large-scale and process-oriented hydro-sedimentological model (WASA-SED) to produce seasonal streamflow and sediment forecasts for the Upper Blue Nile Basin, home to the largest reservoir in Africa (The Grand Ethiopian Renaissance Dam, GERD with a total capacity of 74×109 m3) in Ethiopia. Originating in the Ethiopian highlands, the Blue Nile River, the most important tributary of the Nile, contributes approximately 60% of the Nile’s total flow and is a critical water source for around 20 million people in Ethiopia and 200 million downstream residents in Sudan and Egypt.
WASA-SED was tested and calibrated with river flow data at a daily resolution for the 2001-2007 and validated for 2007-2011. Three different large-scale rainfall “products” were tested and compared ref. their representativity of observed rainfall. We show that such a rainfall evaluation is indispensable for hydrological simulation as well as for seasonal forecasting. We consider this step a “hydrological verification” of rainfall data. Calibration of WASA-SED with river discharge data resulted a Nash–Sutcliffe Efficiency (NSE) of 0.80 and a Relative Error (RE) of 10.32%, while validation results improved to a NSE of 0.81 and a RE of 6.82%.
Seasonal streamflow and sediment flux data were than forecasted for June to December 2024, based on the seasonal meteorological forecast in the preceding month. An ensemble of 51 regional meteorological forecast members in daily resolution and 7 months lead time, each initiating on the first day of each month, was used. Daily and monthly streamflow series were simulated for each forecast member. A post-processing step with an autoregressive model was applied to adjust for forecast biases in seasonal streamflow predictions.
To evaluate the accuracy of 6 months hydrological forecasts, ensemble-averaged monthly rainfall and discharge forecasts were compared with observed average monthly rainfall and discharge values. Results indicate that the coupled meteorological/hydrological models reasonably predict rainfall and discharge on a seasonal scale for the Blue Nile Basin.
The forecasting system is developed in close collaboration with local research partners to facilitate its implementation and sustained use beyond the project's duration.
How to cite: Bronstert, A., Zargar, M., Francke, T., Worku, K., Timale, F., and Kunstmann, H.: Seasonal Streamflow and Sediment Forecast in the Upper Blue Nile Basin, Ethiopia, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6789, https://doi.org/10.5194/egusphere-egu25-6789, 2025.
Water scarcity and drought, worsened by climate change, are growing threats to both the economy and society, particularly in hotspots like the Mediterranean Basin. Effective water resources management requires timely forecasting to implement appropriate countermeasures. Seasonal forecasts, based on knowledge-based models, are essential for managing water scarcity and drought risks. These forecasts have become more reliable with advancements in weather modeling systems. Currently, global seasonal forecasts are provided by various centers, such as the Copernicus Climate Change Service (C3S). However, the success of Artificial Intelligence encourages a shift towards a data-driven approach, which moves away from traditional knowledge-based models, relying instead on machine learning to improve forecast accuracy.
This study compares two advanced seasonal forecast models, one knowledge-based and the other data-driven, for the Calabrian peninsula in southern Italy. The knowledge-based, process-based model uses the SEAS5 ensemble forecasts from the ECMWF, with precipitation predictions disaggregated to a higher resolution (around 9 km) and bias-corrected according to the ERA5-Land product for improved accuracy. The data-driven approach predicts future precipitation using time series from 134 local rain gauges, employing methods like Gaussian process regression (GPR), support vector machines (SVM), and feed-forward neural networks (FFNN). The models’ performance is evaluated for the 2021-2023 period using indices such as bias, RMSE, and Pearson correlation coefficient across different spatial areas. Furthermore, the output of both approaches is used for further hydrological modeling.
The results show a high level of consistency between the two techniques and their respective reference datasets, emphasizing the significant potential of combining both approaches. This integration allows for the utilization of their individual strengths, such as probabilistic forecasting and physical consistency with other variables in knowledge-based methods, as well as flexibility and computational efficiency in data-driven models.
Acknowledgments: This study was funded by The Next Generation EU - Italian NRRP, Mission 4, Component 2, Investment 1.5, call for the creation and strengthening of ‘Innovation Ecosystems’, building ‘Territorial R&D Leaders’, Project Tech4You - Technologies for climate change adaptation and quality of life improvement, n. ECS0000009; The Next Generation EU - Italian NRRP, Mission 4 ‘Education and Research’ - Component C2, Investment 1.1, Research Project of National Interest (PRIN 2022 PNRR) - INnovative FOrecast-informed REServoir operations for sustainable use of water resources and climate change adaptation (INFORES, CUP H53D23001430006), Italian Ministry of University and Research; The Next Generation EU - Italian NRRP, Mission 4, Component 2, Investment 1.3, project WaterWISE - Water Management Strategies and Climate Change Adaptation in Southern Italy, n. PE00000005, CUP D43C22003030002.
How to cite: Senatore, A., Furnari, L., Nikravesh, G., Cortale, F., Lorenz, C., Naghibi, A., Cuartas, A., Kunstmann, H., Bertacchi Uvo, C., and Mendicino, G.: Performance comparison of knowledge-based and data-driven approaches for seasonal meteo-hydrological forecasts in the central Mediterranean, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7157, https://doi.org/10.5194/egusphere-egu25-7157, 2025.
Climate services (CS) play a relevant role in providing tools for establishing societies resilient to global change considering its complex variability at multiple temporal and spatial scales. The involvement of end users in the processes of co-creation, co-development, and co-evaluation of CS, combined with the integration of local data (LD) and knowledge (LK) in forecast modelling and enables the development of user-tailored CS, improving the local impact of climate predictions.
The present work explains the lessons learnt, in terms of CS usability, in the co-creation process developed in seven living labs (LL) within the I-CISK project (https://icisk.eu/). Six LL are located in Europe (Andalucía-Spain, Alazani-Georgia, Budapest-Hungary, Rijnland-The Netherlands, Emilia Romagna-Italy, Crete-Greece) and one in Africa (Lesotho). All of them are identified as climate change hotspots focusing on different climate vulnerabilities affecting different sectors: drought, urban heat waves, water scarcity, landslide susceptibility.
To respond to the different needs and challenges of the 7 LL, we implemented tailored methods for the CS, using LK and LD. These methods include downscaling for seasonal hydrological forecasting, downscaling for meteorological seasonal forecasts and climate projections, seasonal landslide susceptibility forecasts, drought vulnerability assessment and urban heat distribution. In those implementations, LD has a crucial role in all downscaling methods. LK is essential to advisor the selection of explanatory variables into the models, to define alert thresholds in risk events, in the design of adaptation strategies and in supporting vulnerability assessments. On top of these tailored methods, we developed the front-end interfaces of the CS, incorporating user feedback through constant interaction with stakeholders.
From the usability perspective, the main lessons learnt from the experience of the 7 LL in the co-creation of these CS are:
- Most tailored CS aim to produce outputs with higher spatial resolution than those available from existing global, regional, or national services.
- Downscaling techniques are widely applied as tailored methods across many LL, with local data playing a crucial role in these efforts.
- Local data falls short of meeting FAIR data principles.
- Despite the extensive information collected during co-creation processes, only a few LL fully benefit from local knowledge contributions. This knowledge is primarily used to enhance the understanding of climate information, but not to build comprehensive climate knowledge.
- In some LL, using the local language is a requirement for a complete understanding of the climate information, while in others, the scientific/technical terminology poses a barrier.
- Interpretation of the provided climate information (particularly, the uncertainty) is key for developing actions for water resources planning, climate adaptation and vulnerability reduction in different sectors at different LL.
- Sector-tailored information and/or sector-specific indicators are repeatedly demanded in some LL.
How to cite: Pesquer, L., Pechlivanidis, I., Egan, K., Ziogas, A., Mazzoli, P., Castellana, D., Batlle, A., Prat, E., and Bagli, S.: Experiences from seven living labs in the use of local knowledge and local data for tailored climate services, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-8919, https://doi.org/10.5194/egusphere-egu25-8919, 2025.
Lake Urmia, one of the largest saltwater lakes in the world, has experienced a significant decrease in water levels in recent decades. Climate change has been identified as one of the main factors contributing to this reduction. Increases in temperature and changes in precipitation patterns are among the effects of climate change that have led to a decrease in water resources. In this study, five General Circulation Models (GCMs) have been used, including GFDL, MRI, MPI, BCC, and CMCC, to examine precipitation projections in the eastern basin of Lake Urmia under climate scenarios. The precipitation output of the mentioned models has been statistically downscaled using CMHyd software. The selection of GCM models was based on access to their data and their ability to simulate the stations observed precipitation. In the downscaling process, data from three synoptic stations have been used. The observational base period of 1985-2014 and future periods are considered 2026-2050 as near future, 2051-2075 as medium future and 2076-2099 as far future. Also, the future precipitation was predicted under three scenarios: optimistic, moderate and pessimistic, SSP1-2.6, SSP2-4.5 and SSP5-8.5, respectively. The precipitation was studied annually and the changes of precipitation in each of the next three periods compared to the base period of observations and then their significance was examined at a confidence level of 5%. In the optimistic scenario, most models for the periods 2026-2050 and 2051-2075, except for the Tabriz station, indicate a decrease in annual precipitation, and in the period 2076-2099, all stations experience an increase in precipitation which is not significant. In the pessimistic scenario, the BCC, GFDL and CMCC models reveal a decrease in annual precipitation, which in the BCC model is significant for the period 2076-2099. In the middle scenario, except for the GFDL and MRI models which present an increase in precipitation, other models exhibit a decrease in precipitation for all three periods and across all stations. Based on the expected decrease in precipitation in the coming years, and to prevent further drying of Lake Urmia, a national water management program to counteract climate change should be developed.
Keywords: Urmia Lake, Climate change, Precipitation, CMIP6, CMHyd.
How to cite: Pourasghar, F. and Eslahi, M.: Precipitation Projection in Lake Urmia Basin in East Azerbaijan, Iran Using Downscaling of Selected CMIP6 Models, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9450, https://doi.org/10.5194/egusphere-egu25-9450, 2025.
Operational hydrological forecasting involves retrieving, accessing, and processing large volumes of data, alongside managing complex workflows with numerous task dependencies. These challenges are amplified in applications such as flood forecasting, where timely and accurate forecasts are critical for disaster preparedness. Without an efficient workflow manager, significant time is spent diagnosing errors and identifying broken links in the forecasting chain. This inefficiency is particularly problematic in flood forecasting, which demands continuous monitoring and frequent forecast updates—sometimes on an hourly basis—to enable prompt decision-making.
Building on insights from projects such as ULYSSES, we have developed operational hydrological forecasting workflows using pyFlow, a high-level language designed for creating object-oriented suites with ecFlow - the workflow management tool developed by ECMWF. The pyFlow allows users to design, maintain, and execute workflows as software, enhancing efficiency and usability.
In this study, we present the application of pyFlow to develop hydrological forecasting chains that generate ensemble hydrological forecasts on a subseasonal timescale. A key example is the HS2S system, operational since 2021, which provides soil moisture forecasts for Germany using ECMWF ensemble extended forecasts and the mesoscale hydrlogic model (mHM). We detail the transition of workflows from traditional cronjobs to pyFlow on the cluster, showcasing the advantages of this approach.
ecFlow offers a powerful combination of features, including a user-friendly graphical interface, the flexibility to run locally, and open-access customization options. These attributes make PyFlow a versatile tool for both research and operational hydrological forecasting applications. By streamlining workflow management, pyFlow enhances user experiences and supports more effective forecasting and decision-making.
How to cite: Najafi, H., Shrestha, P. K., Kelbling, M., Thober, S., and Samaniego, L.: Streamlining Operational Hydrological Forecasting with pyFlow, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13357, https://doi.org/10.5194/egusphere-egu25-13357, 2025.
In recent years, the Mississippi River has experienced extreme record low water events and flooding. These disruptions have caused significant societal and economic impacts on regional and national scales. Record-setting low water events were recorded in both 2022 and 2023 resulting in unpredicated shipping interruptions throughout the river and saltwater intrusion in southern Louisiana which is dependent on the Mississippi River as a source of drinking water. The National Oceanic and Atmospheric Administration (NOAA) has recognized these record-low water conditions as a climate-related disaster. Volatile conditions on the Mississippi including high impact, high uncertainty low water events are likely to continue to increase with a changing climate. Navigable waterways within the United States are managed and maintained by the US Army Corps of Engineers (USACE). Currently, there is no existing methodology to integrate publically available forecasted riverine conditions with high-fidelity numerical modeling to predict and mitigate economic impacts caused by low water conditions. This information is of direct interest and potential use to stakeholders including the operational departments of individual USACE districts. To fill this gap within the Lower Mississippi river system, a two-dimensional, depth-averaged Adaptive Hydraulics Model of the Lower Mississippi, from Cairo, Illinois to the Gulf of Mexico, was developed. Additionally, a codebase was created to dynamically retrieve 14-day forecasts from NOAA's Office of Water Prediction (OWP) and generate corresponding boundary conditions. This model was validated to the 2023 water year, intermediate forecast points, and observed data during the performance period. The model outputs daily forecasted water levels and depths as spatially referenced rasters and is automated for all steps including boundary condition updates, model runs, and data post-processing. Future work will include the development of dynamically updated bathymetric conditions to reflect shoaling concerns and integration of modeling results into existing USACE GIS based web portal as a decision support tool for stakeholders.
How to cite: Zimmerman, J., Chewning, B., Hammack, A., McAlpin, T., and Jones, K.: Forecast Informed Operational Riverine Modeling: A Case Study on the Lower Mississippi River, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14358, https://doi.org/10.5194/egusphere-egu25-14358, 2025.
The Alpine region is warming faster than the global average, with intensifying heatwaves and declining summer rainfall contributing to increased water shortages and more frequent droughts. Accurate forecasts of meteorological and hydrological variables on a seasonal scale would provide early warnings of extreme seasonal conditions and aid in the management of water resources. The PRIN-2022 SPHERE project1 is developing a forecasting chain based on Copernicus seasonal forecast systems for meteorological inputs and integrates snow-hydrological models to predict seasonal snowpack evolution, river discharge and water availability in the Po River basin (Italy). In this context, it is crucial to assess the skill of Copernicus seasonal forecast systems in predicting meteorological inputs and to evaluate how this skill propagates through the forecasting chain.
This study evaluates the performance of three state-of-the-art seasonal forecast systems available in the Copernicus Climate Change Service (C3S) archive: ECMWF System 5, Météo-France System 6, and CMCC SPS3. These models provide retrospective seasonal forecasts of near-surface air temperature and precipitation at a spatial resolution of 1° x 1° and a monthly temporal resolution, for the common period 1993–2014. The analysis focuses on seasonal average anomalies and applies a range of deterministic and probabilistic verification metrics, including the anomaly correlation coefficient, Brier score, area under the ROC curve, and continuous ranked probability score.
The results provide a comprehensive assessment of the forecast systems' skill in predicting temperature and precipitation anomalies, with a particular focus on the winter and summer seasons-critical periods for applications in energy, water management, agriculture, and the Alpine ski industry. The forecast systems are compared to a baseline forecasting method based on the ERA5 climatology to quantify their relative skills. Preliminary findings reveal strengths and weaknesses across models, with significant variation in performance metrics depending on the season and parameter.
Future steps include extending the analysis to encompass (i) additional meteorological forcings, such as wind and relative humidity, and (ii) output of the forecast chain, including snow water equivalent, snow depth, and river discharge. This will enable the investigation of forecast skill at different steps of the modelling chain and quantify its overall added value compared to a baseline forecasting method based on the climatology. This research aims to advance and optimize the utility of seasonal forecasts in addressing critical climate-related challenges in the Alpine region.
1Progetto di Ricerca di rilevante Interesse Nazionale (PRIN-2022): Seasonal Prediction of wateravailability: enHancing watER sEcurity from high mountains to plains (SPHERE)
How to cite: Pourjavad Shadbad, E., Lorenzo, M., Avanzi, F., Libertino, A., von Hardenberg, J., and Terzago, S.: Assessing the skill of Copernicus seasonal forecast systems in predicting temperature and precipitation anomalies in the Alpine region , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18212, https://doi.org/10.5194/egusphere-egu25-18212, 2025.
Climate and water related disasters impact multiple sectors across a wide range of spatial and temporal scales. There is a need to develop Climate Services (CS) that can meet the short and long-term needs of multiple sectors to build resilience against hydro-climatic disaster risks from droughts and water scarcity, floods, landslides, heatwaves, and windstorms. These risks are projected to increase in the future due to climate change and anthropogenic pressures.
This work presents a case study of co-creation of multi-scalar CS for the Island of Crete, Greece. A co-creation process was designed and executed under I-CISK, an EU funded project (work in progress), to generate climate services that support a multi-sectoral approach towards the tourism sector, by addressing the needs of multiple users and sectors (tourism; water allocation and reservoir management; transportation infrastructure) at the scales which serve both operational (seasonal forecasting) as well as long-term planning needs (decadal projections). Essential to the co-creation process was the bringing together of key players in the CS value chain (providers, purveyors and end users), and the active contribution of a Multi-Actor Platform (MAP), which was composed of members representing policy makers; business and industry; academia and research; and civil society organisations from the sectors addressed.
The co-creation experience revealed a varying nature of needs, perceptions and knowledges attributed to target end-users involved. A combination of joint meetings of the MAP members as well as individual meetings with single sector users was essential in understanding these differences alongside of reducing knowledge and capacity gap among key stakeholders. The role of purveyor (case study leader in this case) was found to be pivotal in holding meaningful exchanges between CS providers and end-users and successfully executing the steps of the co-creation process, such as outlining CS needs/desires and co-designing of the CS. The resulting multi-scalar CS seamlessly integrate global data (e.g., from ECMWF and EU’s Copernicus programme) and local knowledge (e.g., end-user decision-process approach, climate thresholds which trigger responses and historical data) to design CS that combine variables and indicators to cater for the needs of multiple sectors and users across the spatial and temporal scales relevant to them. Moreover, we showcase the salient features of the three co-created CS. Multi-sectoral approach both addresses the complexity of climate change impact on an economic sector as well as increases awareness over the aspects of interrelated impacts and the need of holistic approach towards adaptation planning, among key stakeholders from multiple sectors. The co-created CS demonstrate potential for further development and uptake underpinned by stakeholder’s feedback from application experience in Crete. The findings and experience presented in this work can be instructive for developing multi-scalar climate services in Greece and other countries.
How to cite: Ziogas, A., Masih, I., Tzimas, A., Romas, E., Pechlivanidis, I., Emerton, R., and Werner, M.: Co-creating multi-scalar climate services tailored to the needs of multiple sectors in Crete, Greece, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18742, https://doi.org/10.5194/egusphere-egu25-18742, 2025.
Drought and flood monitoring and forecasting services enable the integration of climate information in decision-making processes, thus supporting successful adaptation, from disaster risk reduction to long-term policy and planning worldwide. While state-of-the-art model-based early warning systems that simulate future floods and droughts at various temporal and spatial scales have been developed in the academic fields, it has proven a challenge to translate the existing knowledge into an operational system. Inherent challenges include a) setting up the appropriate technical requirements, operational workflows, and new IT infrastructure, while b) conducting a user-centric co-development that yields c) scientifically sound products with regard to the model information basis, with d) a continuous financing of operations and scientific updates. However, while transdisciplinary co-development with end users supports utilization, the final provided service will be a balance with respect to the technical and scientific limitations. In particular, technical limitations in the context of a project may lead to less-than-optimal implementation of stakeholder needs, even if those had been scientifically feasible, and therefore may lead to losing actually-elicited, potentially valuable stakeholder knowledge after the process finishes.
This study describes and evaluates the steps and methods undertaken to participatorily co-develop an operational, multi-sectoral global drought hazard forecasting system (in the frame of the OUTLAST project) through a transdisciplinary process of three workshops with participating end users and experts from two focus regions, Lake Victoria Basin, Africa, and Central Asia, respectively. This comprised the co-production of (i) the user-relevant sectoral drought hazard indicators, (ii) the optimal representation with uncertainty information in spatial and temporal visualizations, and (iii) interface functionalities to optimize user utilization of the hazard information. We discuss lessons learned with a particular focus on identified challenges and compromises regarding balancing of the above limitations during the co-development.
The resulting global OUTLAST near real-time monitoring and seasonal forecasts will be operationally provided and freely accessible via the Hydrological Status and Outlook System (HydroSOS) portal hosted by the World Meteorological Organization (WMO). Regarding indicators, we found that the extent of co-design was necessarily limited with a dominating research-lead because of the complexity found in droughts, unlike e.g. in floods. Regarding interface functionalities and user utilization, the technical implementation was limited by sub-optimal funding and a requirement to provide hydrological information homogeneously across all HydroSOS services, and a clear division between an ideal and technically limited version by the IT hosting requirements could be identified. This allowed subsequent planning to provide access to additional features on a potential secondary subportal. We therefore emphasize the importance of 1) eliciting the information on ideal implementation even in the face of current project-bound limitations (i.e., technical or shifting commitment); so that, while 2) promoting wide utilization, the limited functionality can be implemented where public dissemination is most prominent (at WMO level), it is equally important to also 3) plan for alternative approaches to provide more of the ideal features, and to 4) perpetuate the knowledge drawn from the participatory process to ensure the possibility for future implementations beyond the present limitations.
How to cite: Kneier, F., Vu, T., Abbasi, N., Trautmann, T., Weber, J., Dietrich, S., Siebert, S., and Döll, P.: Co-developing a global drought monitoring and forecasting service and lessons learned: eliciting ideal functionalities in the face of real-world implementation limitations, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-20126, https://doi.org/10.5194/egusphere-egu25-20126, 2025.
