Process-based models, such as land surface and hydrologic models (LSMs/HMs), have been foundational to hydrological research for decades. These models are grounded in the principles of mass, energy, and momentum conservation, providing critical insights into the terrestrial water cycle and forming an essential component of Earth System Models. Despite their importance, process-based models face significant limitations, primarily due to parametric and structural uncertainties that hinder their transferability across scales and locations, ultimately reducing their predictive accuracy.
In contrast, machine learning (ML) models learn directly from data, offering potential advantages for capturing highly nonlinear and complex processes, especially when large datasets are available. However, ML models also have notable drawbacks, including a lack of interpretability (often regarded as "black-box" models, despite efforts to develop more explainable or "physically aware" variants), dependence on data quality and availability, and challenges in generalizing under climate or environmental change conditions.
Given the rapid adoption of ML techniques in recent hydrological literature, a key question arises: Can deep learning replace traditional hydrological models due to its speed and accuracy, or is this shift merely a transient trend?
In this presentation, I will argue that before addressing this question, it is essential to establish two key prerequisites: (1) the purpose of the modeling effort, and (2) the appropriate protocols and metrics [1,2] for evaluating model efficiency. To formalize this discussion, I will propose a set of postulations for each modeling paradigm. Drawing on several examples, I will suggest that the most promising future lies in hybrid modeling frameworks, where the empirical aspects of LSMs/HMs (e.g., pedo-transfer function derivation) could be augmented by ML techniques [3,4], while maintaining the core physical processes [5]. ML could also serve as a valuable tool for estimating human-made impacts [6] on the hydrological system, where first-principles models are often lacking.
References:
[1] Rakovec et al. https://doi.org/10.1002/2016WR019430
[2] Samaniego et al. https://doi.org/10.5194/hess-21-4323-2017
[3] Feigl et al. https://doi.org/10.1029/2022WR031966
[4] Li et al. https://doi.org/10.1029/ 2023WR035543
[5] Kholis et al. https://doi.org/10.22541/essoar.173532490.04454195/v1
[6] Shrestha et al. https://doi.org/10.1029/ 2023WR035433
How to cite: Samaniego, L.: Can Deep Learning Revolutionize Hydrology?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-11752, https://doi.org/10.5194/egusphere-egu25-11752, 2025.
The concept of openness (open science, open innovation, open knowledge) has transformed the culture of science and research across the globe, with many scientific disciplines, organizations and countries committing to openness, transparency, accessibility and reproducibility. Development of policy and guidelines further promote Open Science practice by increasing national implementational measures. In this regard, the UNESCO Recommendation on Open Science (UNESCO, 2021) has set an international standard for picking up the pace to evolve together and for each other.
Integration of Open Science into hydrology is gaining higher momentum – there are many exemplary academic initiatives at personal and/or local levels which are scaled up at institutional and/or regional scales. However, there remains a lack of comprehensive strategic framework that advocates for the accessibility of hydrological research to a broad spectrum of researchers, practitioners, and policymakers.
The new “Open Hydrology” publication by UNESCO (https://www.unesco.org/en/articles/open-hydrology) addresses this gap by outlining six pillars— open data, open source, open publishing, open infrastructure, open education, and open participation — to highlight the true potential of Open Science to enhance research transparency, collaboration, and accessibility within water management practices. It is developed for members of (water) research communities and infrastructures, hydrological service providers (including private sector), research administrators and facilitators of research, publishers, policy makers and funders, citizen science groups and initiatives who have a stake in hydrology and water resources research. The key objectives of this publication are:
- to introduce key components of Open Hydrology and discuss required policies, leadership and capacity building,
- to highlight Open Hydrology stakeholders and existing initiatives, tools, resources, etc. for knowledge generation and science governance,
- to establish steps forward on how to address the needs and gaps in implementation of an Open Hydrology framework and,
- to identify opportunities and share recommendations for sustaining Open Hydrology.
In this talk, we will share the highlights from the “Open Hydrology” publication and discuss ways forward to enable the hydrological community to become an ‘Open Science Ambassador’.
How to cite: Dogulu, N., Mertens, A., and Verbist, K.: Elevating Open Hydrology practice and policy: insights for scientists, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-21403, https://doi.org/10.5194/egusphere-egu25-21403, 2025.
Comparing the performance and sensitivity of various hydrological models, or the hypotheses underpinning some of their components present considerable challenges. Biases can be introduced by a multitude of factors, including the expertise or preferences of users and developers, numerical discretisation techniques, input data, preprocessing procedures, calibration strategies, and rounding errors. To minimise these biases, we propose a standardised framework for model comparison and model structure sensitivity analysis.
A flexible hydrological tool, WOLFHydro, has been developed to integrate models organised in modular components. It accommodates models of different natures–with diverse underlying hypotheses– (empirical, conceptual, or physically based) and spatial discretisation approaches (lumped, semi-distributed, or gridded). This tool ensures consistent preprocessing, input data management, semi-distributed catchment discretisation, modelling of anthropogenic structures (e.g., dams and reservoirs) , numerical scheme implementation, and calibration procedures, providing a robust basis for fair inter-model comparisons.
The 2021 floods in most severely impacted Belgian catchments serve as a benchmark case to illustrate the methodology. This study involves comparing hydrological models, which aim to represent the same hydrological processes, but with varying structures, formulations, and nature. It includes an in-house gridded conceptual/physically-based model and widely used lumped models such as GR4H, HBV, NAM, SAC-SMA (Sacramento), and VHM. The proposed framework ensures that the models’ physical outcomes and performance can be compared on equal footing.
This approach not only addresses the issue of equifinality by identifying optimal scenarios but also highlights the strengths and limitations of each model formulation. It emphasises the representation of hydrological processes (runoff coefficient, average contribution of different type of flow in hydrographs, probability of exceedance, etc) over reliance on parameter values alone. This focus would facilitate the parameters transferability of model parameters, particularly in conceptual models where parameters lack explicit physical meaning. Ultimately, this methodology offers a comprehensive framework for improving the transparency, reliability and interpretability of hydrological model comparisons.
How to cite: Dessers, C., Archambeau, P., Dewals, B., Erpicum, S., Borgna, A.-L., and Pirotton, M.: Towards Transparent and Unbiased Hydrological Model Comparisons: A Case Study of the 2021 Floods in Belgium., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18108, https://doi.org/10.5194/egusphere-egu25-18108, 2025.
Event-scale hydrological modeling applications entail fine temporal discretization, enhanced model components, and carefully refined initial and boundary conditions. However, realistic modeling requires justifying assumptions that influence model complexity and the dominant processes represented for a specific catchment. This process is particularly challenging for distributed hydrological models, which, compared to lumped models, incorporate additional assumptions to account for spatial variability in hydrological processes.
This study demonstrates a modeling approach that uses controlled comparisons and meta-metrics of performance to develop a distributed model for a semi-arid catchment in Southern Italy. From hydrological signatures we hypothesize that in this catchment both Hortonian (infiltration excess) and Dunnian (saturation excess) runoff mechanisms can concurrently appear in hydrograph responses during rainfall events. Our objective is to disentangle these mechanisms and design a model capable of distinguishing between them. We therefore developed four perceptual model architectures representing different runoff generation hypotheses, informed by hydrological signatures, and tested them within a nested catchment framework.
A multi-stage operational test involving the calibration of a meta-objective function and spatial transferability validation was conducted to provide a robust and unequivocal ranking of the best-performing models, exposing unsolved structural problems of competing hypotheses. Assessing the consequences of simulated high flows by replacing 2D Shallow Water equations to a simplified routing scheme reinforces the idea of replacing popular metrics with meta-metrics.
Posterior diagnostics confirmed that the most realistic model structure, as indicated by internal consistency in simulated processes, aligned with the highest meta-metrics performance. Hydrographs comparison and hypotheses falsification further revealed that the dominant runoff mechanisms during consecutive storm events could be clearly disentangled, with Hortonian and Dunnian processes alternating depending on rainfall intensity and soil wetness.
By integrating multiple working hypotheses with enhanced operational testing, our proposed model development approach shows that even with limited observational data, such as sole streamflow measurements within a nested catchment setup, it is possible to identify runoff generation processes in event-scale hydrological applications.
How to cite: Perrini, P., Fenicia, F., and Iacobellis, V.: Developing an event-based distributed hydrological model through competing hypotheses and meta-metrics of performance, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-2049, https://doi.org/10.5194/egusphere-egu25-2049, 2025.
Weighted combinations of multiple estimates of the same property is a commonplace technique in hydrology and earth sciences in general. These collections of estimates – ensembles – commonly consist of different realisations of model structures, parametrisations, input data and/or perturbed initial conditions. The use of weighted ensembles is often motivated by their capability to quantify and reduce error and uncertainty. Just how should we derive the weights is not necessarily clear: the literature knows a large number of methods for weighting, ranging from a simple average (the ensemble mean or median) to complex machine learning algorithms, each with various constraints, properties or assumptions. But do the weights derived by different methods make sense? Can we associate the derived weights to the performance of the ensemble members? Are they related to hydrological signatures (hydrological processes)? Do descriptive catchment attributes predict weights associated to certain ensemble members? Understanding these associations is required for appropriate solutions to the major challenge of regionalisation of ensemble weights.
We performed a large sample study of 482 catchments and more than 116 000 simulations of conceptual hydrological model (HBV) and explore how different model averaging methods and constraints to the weights influence the associations and performance of a weighted ensemble. The results show that constraining the weights to strictly positive values is advantageous because the output is less sensitive to the composition and size of the ensemble (i.e. the weights are more stable). Constrained weights do not risk negative streamflow predictions, which can often occur when members are assigned negative weights. Furthermore, constrained weights are more reliable in reproducing flow quantiles (particularly low flows) and flow variation, and their overall performance in the testing period is similar, or better, than predictions derived with weights without constraints. Nevertheless, allowing flexibility of free weights produces outputs with better daily and weekly streamflow dynamics. Based on our explorations on the associations and performance, we present our recommendations for selecting an appropriate model averaging methods in hydrology.
How to cite: Kallio, M., Masoumzadeh, S., and Kummu, M.: On weighted ensembles: do the weights derived with different methods make sense? , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-18145, https://doi.org/10.5194/egusphere-egu25-18145, 2025.
When applying a hydrological model, the length of the calibration period is typically based on the length of the available hydroclimatic data series. Usually, half of the data are used for the calibration of the model and the other half for validation, but other splits (e.g., three quarters and one quarter) are possible as well. When the data record is short, all data may be used for calibration. The general idea is that a longer calibration period will include a wider range of conditions (e.g., a wider range of flood events) and thus lead to a more robust model. However, a longer calibration period does not always have to be better. There are reasons for not using a (too) long calibration period. First, a long calibration period may not be necessary if the extra years of data do not contain any additional information (i.e., different conditions). In this case, a longer calibration period may just waste computer resources, which is an issue when the model is calibrated for a large number of catchments. Second, some discharge records are by now more than 80 years long. During this time period many things have changed. This includes the way that streams are gauged, leading to differences in data accuracy. The catchments themselves will likely have changed as well. For some catchments, these changes are obvious but for other catchments they are more subtle. Even if the dominant land use has remained agriculture, the agricultural practices have changed. Similarly, for catchments that have remained forested during the period of data collection, there may be changes in the percent or spatial pattern of open areas or changes in the species composition. One could, therefore, argue that there is a trade-off between a long calibration period that includes all the variation in the climate and not using data from a period during which the catchment was different from the current conditions. With increasing length of available data series the question on the optimal length of the calibration period becomes more relevant.
To explore the sensitivity of the model results to the length of the calibration period, we calibrated the HBV model for several Japanese and Swiss catchments for which long hydroclimatic records are available. We split these records into multiple calibration and validation periods of different lengths and assessed 1) how the drop in model performance between the calibration and validation period depends on the periods chosen for model calibration and validation, and 2) how the length of the calibration period affects the range in model calibration and validation performances. The results show that the optimal length of the calibration period depends on the catchment, and differs even for neighboring catchments. These analyses provide some information on the optimal length of the calibration period for the study catchments but need to be repeated for other catchments to prove the generalizability of the results.
How to cite: van Meerveld, I., Vis, M., Asano, Y., and Seibert, J.: What is the optimal length of the calibration period?, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13870, https://doi.org/10.5194/egusphere-egu25-13870, 2025.
Mathematical models are idealised representations of real-world processes and must balance the specific characteristics of the object of study with simplifications to ensure usability. In other words: they have to rely on both empirical (backed by data and research) and pragmatic (designed to facilitate computation, abstractions) assumptions. However, we do not know how many of the assumptions embedded in global irrigation models (GIM) fall into each category. Given that pragmatic assumptions are more flexible and can be replaced, changed or removed, this knowledge gap constrains our ability to delineate the uncertainties in these models and assess how reliable their results are.
To tackle this issue, we used sensitivity auditing, a framework for evaluating both quantitative and qualitative assumptions. We systematically analysed 50 documents of nine GIMs and extracted all the assumptions that underpin the simulation of global irrigation water withdrawals. We grouped them into relevant facets of irrigation (climate, crop, soil moisture, irrigation practices, and water source) and classified each assumption as pragmatical or empirical in nature using a philosophy of science perspective.
Our analysis reveals that irrigation models are largely guided by pragmatic considerations. Of approximately 100 identified assumptions, over 70% lack empirical support, with most idealising farmer behaviour. Moreover, 40% of these pragmatic claims are common to at least two models, suggesting that modellers tend to follow each other's assumptions, irrespective of their empirical validity.
The widespread reliance on pragmatic assumptions in GIMs suggests that their uncertainty space is much larger than previously thought, provided that pragmatic assumptions are potentially changeable without jeopardising the representational capacity of the model. The effect that changing pragmatic assumptions has on the output of GIMs deserves further exploration. Our findings underscore the need to appraise the uncertainty in model assumptions to foster transparency and improve the epistemic role and utility of GIMs in society.
How to cite: Linga, S. N., Aguiló, C., Larsen, J., Massimi, M., Wei, N., and Puy, A.: Assumptions in global irrigation modelling are mostly pragmatic, not empirical, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-13900, https://doi.org/10.5194/egusphere-egu25-13900, 2025.
Geoscientific models play a pivotal role in understanding global change impacts on the Earth system and are therefore highly relevant for decision-making. However, their complexity—including the combination of pre-processing, modeling, and post-processing workflows—poses significant challenges to reproducibility and accessibility, even when adhering to FAIR data principles.
Here, we present insights from the land surface modeling community, based on a survey of the 20 dynamic global vegetation models participating in the Global Carbon Project. Our findings reveal substantial room for improvement in software engineering and reproducibility practices and underscore the potential benefits of sharing best practices across modeling communities.
To address these challenges, we highlight tools such as versioning, workflow management systems, containerization, automated documentation, and continuous integration and deployment. These approaches enable reproducible, portable, and automated workflows, ensure code correctness, and facilitate stakeholder access to scientific results.
Finally, we present a showcase of a fully reproducible and portable workflow based on the LPJ-GUESS model, demonstrating how these practices can be implemented and adapted by other modeling communities. This can serve as a resource for improving reproducibility and accessibility, and advancing software engineering standards across geoscientific fields.
How to cite: Gregor, K., Forrest, M., Meyer, B., Darela Filho, J., Schönenberger, U., Justo, V., Bett-Williams, K., and Rammig, A.: Reproducibility and Accessibility in Geoscientific Research: Challenges, Solutions, and Community Perspectives, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-4024, https://doi.org/10.5194/egusphere-egu25-4024, 2025.
Hydrology is a cornerstone for non-real-time flood management decision-making in England, underpinning £6bn of investment by the UK Government. Originally published in 1999, the current prevailing methods used operationally are well-known and resource-efficient but were not designed to address contemporary issues relating to climate and land use changes. It is widely considered that alternative approaches would provide us additional evidence for these issues, but innovation is not cascading into operational practice.
To improve the rate of translation of alternative approached from science into practice, this project, part of the Environment Agency’s (England) Flood Hydrology Improvements Programme (FHIP), will take an existing approach and embed it within operational flood management processes. The journey from science to practice will be documented to better understand the barriers that are faced and how they were overcome, looking wider than simply method development to consider ‘quality-of-life’ factors (e.g. user interfaces) and training.
This presentation will showcase the discovery phase of the project. This includes research into: what the user needs and requirements are; the blockages to methods making the leap from science to practice; what can we learn from international practice; what are the best ways to communication uncertainty; and what information about climate change impacts do we need to capture for decision-makers. Future plans will be outlined for the project, including the development of new and novel open-source software to encourage reporting of decision-points and uncertainty in the modelling process.
How to cite: Skinner, C. and Asadullah, A.: Science to Practice: Embedding new hydrology approaches for flood management decision-making., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6993, https://doi.org/10.5194/egusphere-egu25-6993, 2025.
Hydrology faces critical challenges in reproducibility, accessibility, and collaboration, limiting progress and innovation. We introduce “Moving Research Down the Academic Career Scale” (MRDACS): the idea that work should be reproducible by someone at an earlier career stage and in less time than the original work. We advocate for research tools and methods to be accessible to students and early-career researchers. By embedding Open and FAIR (Findable, Accessible, Interoperable, Reusable) principles, modular tool design, and user-friendly interfaces, we can lower barriers to reproducibility and foster equitable participation in hydrological research. We will showcase practical strategies to empower researchers at all levels to build on existing work, reducing time spent overcoming technical challenges and enabling deeper focus on innovation. We present our, and our students, science done over the last decade on the eWaterCycle platform to illustrate how we have practically implemented Open and FAIR principles to support MRDACS. This approach advances equity and inclusivity while strengthening collaboration across academic and professional communities. By prioritising reproducibility and transparency, we can create a more resilient and impactful hydrological science field equipped to tackle urgent global challenges.
At the time of abstract submission, this work has been submitted to, and is in review in, Philosophical Transactions A.
How to cite: Hut, R. and Hall, C.: Moving Research Down the Academic Career Scale (MRDACS), EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7986, https://doi.org/10.5194/egusphere-egu25-7986, 2025.
Hydrological models play a crucial role in understanding and managing water resources. However, accurately representing complex streamflow generation processes remains a significant challenge. We introduce an innovative diagnostic framework designed to evaluate process limitations in hydrological models, emphasizing event-based and multi-dimensional assessments. The framework first evaluates model error variability by classifying streamflow events into distinct types (e.g., Snow-or-Ice, Rain-on-Dry, Rain-on-Wet) and leveraging multi-dimensional metrics (i.e., timing and relative magnitude errors). It then assesses the importance of error drivers (e.g., hydrographic properties, model fluxes and states, model inputs, and pre-event errors) using explainable machine learning (XAI). A case study involving 340 German catchments demonstrates the framework's applicability. The results reveal that the majority of model-simulated streamflow events exhibited time delays and magnitude underestimations. Specifically, Rain-on-Dry events showed higher timing errors, while Snow-or-Ice events had larger relative magnitude errors. Furthermore, errors varied across different hydrograph components (pre-event, rising limbs, peaks, and recession limbs) for each event type. Simulated streamflow at all components, especially peaks, was predominantly delayed in timing and underestimated in magnitude in more than 50% of events. Using Random Forest regression with Accumulated Local Effects, the analysis found that pre-event errors are the dominant driver for both timing and relative magnitude errors across all event types. The relative magnitude errors were also strongly affected by hydrograph-related event properties and model fluxes and states for land surface and groundwater dynamics, with these drivers having greater importance for Snow-or-Ice events. This framework enhances diagnostic capabilities, providing a robust tool for advancing hydrological model evaluation and understanding under diverse hydrometeorological conditions.
How to cite: Tarasova, L., Wang, Z., and Merz, R.: From Events to Insights: Event-Process Based Diagnostics of Hydrological Model Performance, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6927, https://doi.org/10.5194/egusphere-egu25-6927, 2025.
The ratio of uncertainty to mutual information (RUMI) is proposed as a new and novel objective function for rainfall-runoff model calibration. Uncertainty is quantified by means of BLUECAT (likelihood-free approach), whereas mutual information through entropy-based concepts. The deterministic GR4J rainfall-runoff model is considered to illustrate RUMI’s calibration capabilities over around 100 catchments in Chile. Those catchments have a pseudo-natural hydrological regime and are located in different macroclimatic zones. Calibration with the Kling-Gupta Efficiency (KGE) was also performed. Additionally, several hydrological signatures were used to assess RUMI’s performance and comparison with KGE-based results was carried out. Key findings showed that RUMI-based simulations had improved performance and reduced variability (in comparison with KGE-based simulations). This study highlights RUMI’s capabilities for hydrological model calibration by considering uncertainty quantification as a key computation step and, therefore, contributing to more accurate and reliable hydrological predictions. This work was supported by The National Research and Development Agency of the Chilean Ministry of Science, Technology, Knowledge and Innovation (ANID), grant no. FONDECYT Iniciación 11240171; the RETURN Extended Partnership which received funding from the European Union Next-GenerationEU (National Recovery and Resilience Plan – NRRP, Mission 4, Component 2, Investment 1.3 – D.D. 1243 2/8/2022, PE0000005); and, the Italian Science Fund through the project "Stochastic amplification of climate change into floods and droughts change (CO$_2$2Water)", grant number J53C23003860001.
How to cite: Pizarro, A., Koutsoyiannis, D., and Montanari, A.: RUMI (Ratio of Uncertainty to Mutial Information): Uncertainty consideration in rainfall-runoff models calibration, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12388, https://doi.org/10.5194/egusphere-egu25-12388, 2025.
Dams and reservoirs are integral to regional water management, providing critical services such as flood and drought control, water supply, hydropower generation, and recreation. However, their streamflow regulation often disrupts hydrological connectivity, sediment transport, and biodiversity, leading to significant ecological consequences. These alterations modify flow regimes across various time scales (hourly to annual), complicating the accuracy of hydrological models in affected regions. Thus, understanding how dam-induced streamflow alterations propagate through river networks is essential for informed water resource management.
Current flow regulation indicators, such as those official flags from Water Survey Canada (WSC), are point-scale binary values that often under- or over-estimate regulation effects and lack spatial continuity. To address this limitation, we propose a spatially continuous metric, the Streamflow Alteration Index (SAI), which incorporates point-based alteration signals from dams, reservoirs, lakes, hydropower facilities, and hydrometric gauges into a subbasin-scale river and routing network. The SAI allows hydrologists to quantify cumulative upstream streamflow alterations at any point in a vector-based routing network. Using Ontario, Canada, as a case study, we applied the SAI to a network encompassing 245,576 subbasins, 82,928 lakes, and over 3,000 alteration sources identified from provincial and global datasets. This approach produced a seamless, high-resolution map of streamflow alteration signals across Ontario (total area: 1.07 million km2) at the subbasin scale, importantly covering both gauged points (including 1,320 flow and level gauges) and ungauged locations within the routing network. The SAI was validated against nearly 500 hydrometric gauges with WSC regulation flags.
Results demonstrate that the SAI effectively identifies near-natural gauges with over 95% accuracy while revealing that more than 40% of gauges that are flagged as regulated by WSC could instead be reconsidered as model calibration targets, as many of them show little signs of significant regulation. By offering a less restrictive yet more reliable alternative, the SAI enables hydrologists to retain a larger pool of near-natural gauges for calibration, thereby enhancing streamflow predictions, particularly in data-sparse or ungauged regions. Furthermore, the SAI approach can be generalized to other routing networks in Canada and globally.
How to cite: Shen, H., Tolson, B., Craig, J., Metcalfe, R., and Romero Cuellar, J.: Streamflow Alteration Index (SAI): Mapping Dam and Reservoir Impacts on Streamflow for Improved Hydrological Modeling, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12698, https://doi.org/10.5194/egusphere-egu25-12698, 2025.
Good representation of the hydrological system in models is required to provide reliable predictions. The selection of a suitable set of performance criteria is a core decision in identifying the optimal parameter set(s) during model calibration. As each performance criterion focuses on different parts of the hydrograph, their selection often determines which parameter values are selected as optimal for representing the rainfall-runoff behaviour in a catchment. Knowning which performance criteria are most suitable for which purpose, model or catchment is difficult to determine.
We therefore selected a set of 16 classical performance metrics and signature measures which together cover all phases of the hydrograph to test their suitability for identifying different types of parameters. We used four hydrological models (HBV, SWAT+, mHM and RAVEN-GR4J) in six catchments belonging to diverse landscapes in Germany. All model parameters were grouped into five process groups (snow, evapotranspiration, soil, surface and subsurface processes) to make the parameters comparable between the models. We then developed a metric called “identifiability quote index” which shows the degree of identifiability for each combination of parameter and performance criterion.
Our results show that the classical performance criteria (e.g. NSE, KGE) are not sufficient to identify suitable values for all parameters. Signature measures (e.g. flashiness index, baseflow index) often have a higher “identifiability quote index” for specific cases and are suitable for either capacity or flux parameters. The degree of identifiability tends to vary between processes and models, but evapotranspiration parameters are generally highly identifiable with water-balance related metrics. The more complex a model is (e.g. mHM, SWAT+), the more difficult it is to determine parameter identifiabilities.
In conclusion our study shows that a set of contrasting performance metrics and signature measures are needed to represent the whole hydrological system and to accurately identify the parameters.
How to cite: Guse, B., Herzog, A., Houska, T., Spieler, D., Staudinger, M., Wagner, P., Thober, S., Loritz, R., and Pool, S. and the DFG Scientific Network IMPRO: Evaluation the suitability of contrasting performance metrics and signature measures with the identifiability quote index, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14940, https://doi.org/10.5194/egusphere-egu25-14940, 2025.
Predicting low-flow characteristics in ungauged basins is crucial for effective water management. Regionalization of low flow can either directly focus on lumped characteristics, such as mean annual minimum (MAM), or on seasonal or monthly characteristics (e.g., mean winter minima, mean summer minima, monthly mean minima). Alternatively, regionalization can focus on the time series (e.g., annual, monthly, or daily time series), as in rainfall-runoff models, which are subsequently used to predict the characteristics of interest. Most studies to date have regionalized runoff characteristics separately, leading to inconsistencies for each catchment. We propose regionalizing a full time series for each site to derive all low-flow characteristics from this single time series.
We regionalize daily streamflow and monthly, seasonal, and lumped low-flow characteristics using the US-CAMELS dataset. Low-flow characteristics are derived from the 7-day average streamflow, allowing us to compare annual, seasonal, and overall minima across different regionalization methods. Our approach leverages state-of-the-art machine learning models, such as tree-based models, support vector regression, and deep-learning architectures. For rainfall-runoff modeling of daily streamflow, we use an LSTM model tailored to low-flow prediction with an expectile loss function. Model validation is performed using 10-fold cross-validation. We evaluate our approach not only with common error metrics - such as RMSE and MAE - but also by quantifying the error in estimating the extreme value distribution of annual minima from the predicted time series.
Our results indicate that higher temporal resolution yields higher prediction accuracy compared to lumped characteristics. However, tailoring daily streamflow predictions to the lower quantile of the data is essential for more accurate results.
How to cite: Laimighofer, J., Bachler, A., and Laaha, G.: What should we actually regionalize? - The benefits of temporal aggregation for low-flow prediction., EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-19344, https://doi.org/10.5194/egusphere-egu25-19344, 2025.
Modeling evapotranspiration is an increasingly relevant topic in scientific discussions, as its volume and trends are essential for identifying climate change. However, there is still no accepted method use as a reference for evapotranspiration modeling. The most preferred method is the FAO65 Penman-Monteith (P-M) model, which is widely used as a reference method for calculating reference and crop evapotranspiration and is recommended by scientific authorities. The aim of the research was to regionalize the Hargreaves model for calculating reference evapotranspiration under Central European conditions, aiming to achieve accuracy as close as possible to the P-M model.
A significant finding of the study is that the model coefficients are not stable over time, and therefore the accuracy of any modification to the Hargreaves model must be regularly validated. Our results revealed a decrease in the accuracy of the modified Hargreaves model as the altitude of the climatological station increased. When altitude was incorporated into the Hargreaves equation, the model's accuracy significantly improved for stations at higher elevations, achieving a consistent level of accuracy across all stations, regardless of their location or altitude. Additionally, the results suggested that the optimal values for the model coefficients vary over time, with the B coefficient showing a decreasing trend of -0.5 and the C coefficient declining by -0.1 between the periods 1981-2000 and 2001-2020. This issue is particularly pronounced in the analysis of shorter time periods, where model may lead to substantial accuracy reductions.
How to cite: Rattayová, V. and Garaj, M.: Assessing Changes in Hargreaves Evapotranspiration Model Accuracy Across Time and Altitude, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9601, https://doi.org/10.5194/egusphere-egu25-9601, 2025.
Practical application of some new technologies, and the light they shed on model design and model problem decomposition
New robust efficient modelling technologies are available to quantify the uncertainties associated with model predictions. We adopt a decision support modelling framework which uses a combination of two of these new technologies, Data Space Inversion and Ensemble Space Inversion. Using this framework helps answer model design and deployment questions that are critical for a specified decision. These questions include:
- What contributes to the uncertainty of what could go wrong with this decision?
- Where is the information that may reduce this uncertainty?
- How can this information be best harvested – what model structure, parameterisation, observation weighting strategy, and technologies are most appropriate?
- How are the consequences of information insufficiency best expressed?
We demonstrate how this modelling framework reveals the predictive accuracy costs of over-fitting to some types of data. We also identify for a specific prediction which alternative model structures and inversion methods are more appropriate given alternative data sets.
How to cite: Moore, C., Kitlasten, W., and Doherty, J.: Practical application of some new technologies, and the light they shed on model design and model problem decomposition, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-4135, https://doi.org/10.5194/egusphere-egu25-4135, 2025.
The siting process of a deep geological nuclear repository is a complex long-term endeavour involving many different stakeholders. Assessing the
suitability of a site for a nuclear waste repository requires, among others, robust simulation models of the relevant underground thermal, hydrological,
mechanical, and chemical (THMC) processes. Screening such sites for an entire country involves running these simulation models for various parameter
sets, on various scales, with various degrees of simplification. The data integration from different sources and post-processing and visualization of
results are of equal importance as the models themselves.
The OpenWorkFlow project, funded by the Bundesgesellschaft für Endlagerung (BGE), aims at developing open source, automated, robust, quality assured simulation workflows in the context of the nuclear waste repository siting process in Germany. During the first project phase from 2021 to 2024 several software products emerged from the OpenWorkFlow project, which will be presented on this poster:
- OGSTools, a Python tool suite around OpenGeoSys (OGS), the reference THMC simulator in the OpenWorkFlow project, simplifying model setup, simulation studies and post-processing (https://ogstools.opengeosys.org).
- A FEFLOW to OGS converter, enabling to combine the advantages of both simulators: the convenience and UI features of FEFLOW and the transparency and extensibility of OGS (https://ogstools.opengeosys.org/stable/user-guide/feflowlib.html).
- A fully automated workflow for the thermal dimensioning of a nuclear waste repository—i.e., determining the required area for a repository—for various parameter combinations. This workflow has already been used in the siting process in Germany in practice.
- A set of virtual reality applications have been developed that provide a virtual field trip to the Mont Terri rock laboratory and a serious game in an immersive virtual environment. These applications support the exploration and validation of underground processes. They enable an improved science communication of conducted research as well as training and collaboration on the ongoing experiments.
The second phase of the OpenWorkFlow project has started in January 2025. Until the end of 2029 the existing simulation workflows will be heavily extended to support the safety assessment of nuclear waste repository candidates.
How to cite: Lehmann, C., Bilke, L., Graebling, N., Heinze, J., Meisel, T., Naumov, D., Sen, Ö. O., and Kolditz, O.: Software Products from the OpenWorkFlow Project, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-15544, https://doi.org/10.5194/egusphere-egu25-15544, 2025.
Earth system models are able to simulate the physical processes that govern the Earth's climate system and are essential to understand and predict climate change. However, these models come at a significant computational cost since they need to simulate a multitude of variables at a high temporal and spatial resolution to adequately represent the climate system. Climate model emulators are statistical models that are trained to reproduce (emulate) selected variables of full-fledged physical climate models at a much lower computational cost and higher speed. Such emulators are especially interesting in the context of climate change mitigation policies, which often deal with a limited number of relevant variables (e.g. annual mean temperature, number of hot days, annual maximum precipitation, etc.), but require several scenarios of how these variables may evolve under different policy choices. The “Modular Earth System Model Emulator with spatially Resolved output”, in short MESMER, is a climate model emulator that can emulate large ensembles of several climate variables (see list of modules below) for any future climate change scenario conditional on global mean temperature.
Four MESMER modules have been developed over the last five years by several researchers from different institutions: (1) MESMER: module for annual mean temperature, (2) MESMER-M: module for monthly mean temperature, (3) MESMER-M-TP: module for monthly mean temperature and precipitation, and (4) MESMER-X: module for conditional distributions with focus on climate extremes. These modules were developed largely independent of each other and grew organically to meet the needs of the individual researchers and the analyses they performed without following consistent coding standards or software architecture.
Here we present how we unified the MESMER code base, integrating all modules into a single repository and rewriting them to adhere to sustainable software standards. We redesigned MESMER with respect to (1) maintainability, (2) extensibility, (3) flexibility, (4) adherence to a defined software architecture and, (5) accessibility. The result is an open source software tool that anyone can use and/or extend. Moreover, the software is easily available and understandable to users who are interested in emulating variables for their own scenarios without being proficient in climate modelling, for example policy makers. In addition, MESMER output from the revised modules is stable and reproducible. Here we present the unified MESMER version 1.0.0 and provide insights into the achievements, challenges and lessons learned during this process. This includes insights into the chosen architecture, our testing and code review framework, stability and performance enhancements and recommendations for the scientific programming community.
How to cite: Bauer, V., Hauser, M., Quilcaille, Y., Schöngart, S., Gudmundsson, L., and Seneviratne, S.: Presenting MESMER v1 - Integrating Multiple Climate Emulator Modules Into One Sustainable Research Software Package, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9493, https://doi.org/10.5194/egusphere-egu25-9493, 2025.
We present a new coupling framework named FINAM (short for "FINAM Is Not A Model"). FINAM is designed to facilitate the coupling of models that were developed as stand-alone tools in the first place, and to enable seamless model extensions by wrapping existing models into components with well-specified interfaces. Although established coupling solutions such as YAC (Hanke et al., 2016), ESMF (Collins et al., 2005), or OASIS (Craig et al., 2017) focus on highly parallel workflows, complex data processing, and regridding, FINAM prioritizes usability and flexibility, allowing users to focus on scientific exploration of coupling scenarios rather than technical complexities. FINAM emphasizes ease of use for end users to create, run, and modify model couplings, as well as for model developers to create and maintain components for their models. The framework is particularly suited for applications where rapid prototyping and flexible model extensions are desired. It is primarily targeting environmental models, including ecological models for animal populations, individual-based forest models, field-scale crop models, economical models, as well as hydrologic and hydrogeological models. Python's robust interoperability features further enhance FINAM's capabilities, allowing to wrap and use models written in various programming languages like Fortran, C, C++, Rust, and others. We will describe the main principles and modules of FINAM and presents example workflows to demonstrate its features. These examples range from simple toy models to well-established models like OpenGeoSys and Bodium covering features like bidirectional dependencies, complex model coupling, and spatio-temporal regridding.
Links
- FINAM website: https://finam.pages.ufz.de
- FINAM paper preprint: https://doi.org/10.5194/gmd-2024-144
Refrences
- Hanke, M., Redler, R., Holfeld, T., and Yastremsky, M.: YAC 1.2.0: new aspects for coupling software in Earth system modelling, Geosci-
entific Model Development, 9, 2755–2769, https://doi.org/10.5194/gmd-9-2755-2016, publisher: Copernicus GmbH, 2016. - Collins, N., Theurich, G., DeLuca, C., Suarez, M., Trayanov, A., Balaji, V., Li, P., Yang, W., Hill, C., and da Silva, A.: Design and Implemen-
tation of Components in the Earth System Modeling Framework, The International Journal of High Performance Computing Applications,
19, 341–350, https://doi.org/10.1177/1094342005056120, publisher: SAGE Publications Ltd STM, 2005. - Craig, A., Valcke, S., and Coquart, L.: Development and performance of a new version of the OASIS coupler, OASIS3-MCT_3.0, Geoscien-
tific Model Development, 10, 3297–3308, https://doi.org/10.5194/gmd-10-3297-2017, publisher: Copernicus GmbH, 2017.
How to cite: Müller, S., Lange, M., Fischer, T., König, S., Kelbling, M., Rojas, J. J. L., and Thober, S.: FINAM - is not a model (v1.0): a new Python-based model coupling framework, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-6421, https://doi.org/10.5194/egusphere-egu25-6421, 2025.
Located at the interface between the groundwater table and the atmosphere, soil lies at the core of the critical zone. It is a complex, dynamic environment sustaining essential ecosystemic services and biodiversity. Numerical simulation models of soil processes are invaluable tools for tackling the complex issues involved in understanding and predicting physical, chemical and biological cycles, in relation to agricultural production, soil protection and adaptation to climate change. To provide a detailed representation of soil functioning, it is necessary to couple a large number of models that represent the various processes taking place within it. Modelling platforms help to do this by facilitating the development and use of coupled models of soil processes. A key requirement of such platforms is to be able to integrate existing, already validated, models without major difficulties.
To this aim, we present the VSoil modelling software platform (https://vsoil.hub.inrae.fr/) developed at INRAE (France’s National Research Institute for Agriculture, Food and Environment) since 2009 in close collaboration between scientists and software engineers. VSoil is an open-source platform designed to aid the development of numerical models at the soil profile scale describing physical, chemical and biological processes in soil and its interactions with climate and plants but also anthropic activities. The user-friendly workflow of VSoil simplifies the development and use of models, making them accessible even to scientists with limited experience in computer programming. The VSoil software suite comes with a range of already developed models and is designed to guide users as much as possible in addressing their scientific questions, by providing tools for: i) defining and describing pertinent soil processes and their interactions through their input and output variables, ii) developing elementary models, called modules, which are numerical representations of the processes, iii) assembling and coupling these modules into more or less complex models, and iv) parametrising and executing the resulting models, and visualising results. The VSoil team provides user support and regularly adds new features to meet the needs of the user community. VSoil currently offers key features, including: i) model exploration tools (sensitivity analysis and parameter estimation) along with the ability to run models on several sets of input data, ii) the possibility to run models, in a reproducible way, on a remote computing environment (server or cluster), iii) the connection to INRAE's national agroclimatic database. VSoil fosters collaboration between scientists from various disciplines and facilitates the sharing and use of new developments within the platform's user community.
VSoil is being used by scientists from various countries to address very diverse questions such as the fate of persistent fluorinated pollutants in soils, the impact of treated wastewater on soil, the use of geophysics for non-destructive characterisation of soil hydraulic properties, the fate of pesticides at the landscape level, the simulation of soil carbon dynamics, or the optimisation of forestry machinery operations to mitigate soil degradation and compaction.
How to cite: Beudez, N., Moitrier, N., Moitrier, N., Nouguier, C., Ruy, S., and Lafolie, F.: Coupling easily numerical models using the VSoil modelling platform, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9899, https://doi.org/10.5194/egusphere-egu25-9899, 2025.
Hydrological modeling frequently requires time-intensive preprocessing of spatial datasets, including Digital Elevation Models (DEMs), to provide inputs in the desired format for the chosen hydrological model. It takes up important research time and is difficult to repeat. V-FOR-WaTer as a virtual research environment provides a practical methodology for environmental data processing, equipping researchers with tools that streamline model construction, improve reproducibility, and minimize errors.
This abstract emphasizes a use case involving V-FOR-WaTer’s GIS preprocessing tools, which are intended to automate the creation of hillslope geometry files for the spatially distributed hillslope-scale hydrological model CATFLOW. The automated operations convert raw DEMs into key input files, including streams, flow accumulation, aspect, distance to rivers, and elevation profiles. These instruments diminish manual preprocessing duration and enhance repeatability, allowing researchers to concentrate on analysis and scenario formulation.
The GIS preprocessing tool is developed in line with FAIR (Findable, Accessible, Interoperable, and Reproducible) principles, enhancing their adaptability to diverse regions and datasets. Their standardization and accessibility enable seamless integration into various research workflows, fostering consistency and scalability. While the full workflow is under development, preliminary results demonstrate the platform’s potential to harmonize datasets and improve hydrological modeling efficiency.
V-FOR-WaTer encourages hydrologists and environmental scientists to examine the CATFLOW workflow and use its tools to enhance efficiency and reproducibility in hydrological research. This approach ensures that researchers can obtain reliable results while reducing the difficulties associated with manual data preparation.
How to cite: Bouguezzi, S., Azmi, E., Bishop, B., Boussaoud, K., Dolich, A., Hassler, S. K., Mälicke, M., Manoj Jaseetha, A., Meyer, J., Streit, A., and Zehe, E.: Automating Hydrological Model Preprocessing: GIS Tools for CATFLOW in V-FOR-WaTer, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-14083, https://doi.org/10.5194/egusphere-egu25-14083, 2025.
This research focused on the application of Python-based tools for efficient preparation and processing of input data for hydrological modelling in an agricultural catchment of the Kocinka River (SW Poland). Prepared scripts and workflows address the challenge of integration of many data sources required for SWAT+ and MODFLOW models. The presented study focuses on automation of data preprocessing tasks and model calibration support, with option to reuse scripts in future work with similar data for different areas.
The Python-based approach utilizes various libraries, like: GeoPandas for processing spatial data from vector maps, Pandas and Numpy for handling meteorological time series from the Polish Institute of Meteorology and Water Management (IMGW), and Flopy for MODFLOW data management. The scripts streamline the preparation of weather and soil input data specifically formatted for SWAT+ Editor and QSWAT, significantly reducing manual data handling and potential errors in the data preparation phase. The automated workflow particularly benefits the processing of data from agricultural areas, which comprise 66% of the catchment area, ensuring consistent handling of land use parameters across the modeling domain.
The data processing framework incorporates multiple data inputs: meteorological data including precipitation, temperature, and other climate variables, detailed soil maps and land use information as well as satellite data about solar radiation (SARAH-2). The system processes river stage data from three profiles with 30-minute temporal resolution, complemented by flow measurements for hydrological validation.
The developed Python tools also support the model calibration process by enabling rapid modification of input parameters and automated analysis of water balance components. This approach allows for efficient sensitivity analysis and model refinement, particularly beneficial for understanding the groundwater-surface water interactions.
The study contributes to good modeling practices by providing examples of efficient data preprocessing workflows and calibration support tools, essential for complex hydrological studies that combine multiple data sources and modeling platforms. The automated approach not only saves time but also enhances reproducibility and transparency in the modeling process.
Acknowledgements. The work was carried out as part of WATERLINE project (2020/02/Y/ST10/00065), under the CHISTERA IV programme of the EU Horizon 2020 (grant no. 857925) funded by National Science Centre, Poland and a partially by AGH University of Krakow, Faculty of Geology, Geophysics and Environmental Protection (grant no. 16.16.140.315).
How to cite: Nikiel, M., Szymkiewicz, A., Wachniew, P., and Żurek, A. J.: Enhancing data preparation for hydrological modeling: a Python-based approach for coupling SWAT+ and MODFLOW, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12133, https://doi.org/10.5194/egusphere-egu25-12133, 2025.
The use of deep geothermal energy is an important building block in the planned transformation of energy systems. However, the development of new sources, particularly for the pore aquifers in the North German Basin, has not progressed as far as necessary to make a substantial contribution. Decisive obstacles here are, on the one hand, the exploration risk, which essentially results from uncertainties regarding the subsurface properties, as well as larger cost factors that are difficult to calculate during operation, such as scaling effects, which reduce efficiency and may require expensive countermeasures. The lack of information on the geological properties in the target horizon makes it difficult to plan and make the estimates required for decision-making. Due to the depth of the formations, the generally weak information and data situation will not change quickly. In addition to the further development of exploration methods, methods are therefore needed that generate the best possible information about the subsurface and the processes from the available data and take into account the uncertainties of the information obtained due to the limited data available.
In this contribution we will present the approaches of a project where this question will be tackled by developing AI methods for accessing and linking existing data sources. The aim of the project is to develop and apply an IT-based concept for the planning of geothermal duplicate systems in northern German aquifers and to predict the influence of geochemical processes on the long-term efficiency of these systems. For this purpose, a digital twin of the subsurface with an assessment of uncertainties is being developed and various coordinated digital tools are being created and combined in an open-source workflow that can be flexibly modified. This is being developed as an example for an existing geothermal power plant, which has been in operation for many years; in particular the geochemical processes that have been investigated for the plant for a long time are being taken into account. The results of the project are intended to support planning and decision-making and make existing process and site knowledge available for more efficient operation of deep geothermal energy.
The presentation will give an overview of the project and present initial work.
How to cite: Sester, M., Neuweiler, I., Broggi, M., Stechern, A., and Martin, T.: ThermoOptiPlan: Optimizing Planning and Operation of Geothermal Systems using innovative prediction tools, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-12401, https://doi.org/10.5194/egusphere-egu25-12401, 2025.
