Stochastic Hydrology with contributions on methodologies and applications, for modeling, forecasting, change assessment, and uncertainty quantification

Stochastic hydrology offers efficient tools for characterizing processes in hydroclimatic systems, e.g., for hydrologic design, hydroclimatic systems modeling and forecasting, and water resources management. Theory and application of stochastic processes enables a faithful and consistent representation of natural processes that in many cases outperforms outcomes of physically based models. Stochastic modelling offers the means to mimic the variability of processes in space and time, and to characterize the inherent uncertainty in probabilistic terms. For example, this allows to simulate synthetic space-time fields reproducing the characteristics of the process – the main statistical properties across multiple spatial and temporal scales – for assessing the hydrological impact in a complex and changing environment.
This session calls for papers developing and discussing stochastics tools to systematically deal with uncertainty, constant or sudden change, and space-time variability, for characterization or simulation (including disaggregation) purposes of hydroclimatic variables such as precipitation, temperature, streamflow, or soil properties. Contributions are invited, for instance, on the improvement of stochastic modeling in hydrology, innovative techniques for identifying model structure, calibrating parameters, assessing uncertainties, etc. (see also the Unsolved Problems in Hydrology UPHs 1-4 and 5-8 identified by Blöschl et al., 2019, which are related to time-variability and change and space-variability and scaling, respectively).

Convener: Elena Volpi | Co-Conveners: Simon Michael Papalexiou, Antonio Zarlenga, Marco Marani, Alberto Viglione
| Wed, 01 Jun, 08:30–15:00|Room Barthez 1
| Attendance Wed, 01 Jun, 15:00–16:30|Poster area

Orals: Wed, 01 Jun | Room Barthez 1

Chairperson: Elena Volpi
Svenja Fischer and Andreas Schumann

The regionalisation of flood frequencies is a precondition for the estimation of flood statistics for ungauged basins. We propose a flood type-specific regionalization, to take into account the different flood-generating processes. The different flood types are classified according to their meteorological causes. Their probability distributions are modelled by type-specific distribution functions which are combined into one statistical annual mixture model. For regionalisation, we specified the parameters of each type-specific probability distribution separately with hierarchical clustering and regressions from catchment attributes. By selection of the most relevant features, depending on the flood type, the specifics of flood-generating processes were considered. This approach offers a higher degree of freedom for regionalisation as it describes the relationships between catchment characteristics, meteorological causes of floods and response of watersheds. However, as for many regionalisation procedures, uncertainty is created by different observation periods of the gauged catchments used for regionalisation. Due to different observation periods, extreme events can be included in one sample, while for the neighboring catchment with shorter observation period this may not be the case. This stochastic uncertainty results in a heterogeneity of second and third moments which may bias the regionalization. We analyzed different scenarios to take into account this uncertainty and to include the information of long observation periods as well as the occurrence of extreme events for correcting the distribution parameters accordingly. This leads to more homogeneous regionalisation results.

How to cite: Fischer, S. and Schumann, A.: Consideration of sample size uncertainty for the regionalization of type-specific flood frequency analyses, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-58, 2022.

Matteo Pesce, Alberto Viglione, Jost Hardenberg, Larisa Tarasova, Stefano Basso, and Ralf Merz

Large scale modelling is becoming increasingly important in hydrology, particularly to characterize and quantify changes in the hydrological regime, whose drivers are typically large-scale phenomena, up to the global scale (e.g., climate change). This can be done with distributed models by estimating spatially consistent model parameters, that are parameters having a functional relationship with catchment characteristics. In this study we adopt the newly developed PArameter Set Shuffling (PASS) approach, based on a machine learning decision tree algorithm, for the regional calibration of the SALTO (SAme Like The Others) model. The method exploits observed patterns of locally calibrated parameters and catchment (climatic and geomorphological) descriptors, to derive functional relationships between the variables. The results, for around 100 catchments located in North-Western Italy, demonstrate that the use of regionally calibrated parameter sets results in similarly good performances as for local calibration. This suggests that the predicted parameter sets can be efficiently used for streamflow prediction in ungauged basins or under changing conditions.

How to cite: Pesce, M., Viglione, A., Hardenberg, J., Tarasova, L., Basso, S., and Merz, R.: How much value is in climate and geomorphological descriptors for distributed hydrological modelling? A decision tree based approach in North-Western Italy, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-361, 2022.

Andrea Magnini, Michele Lombardi, Elena Valtancoli, and Attilio Castellarin

Due to the limited length of locally available sequences of precipitation extremes, point rainfall depth associated with given duration and return period is usually estimated through regional frequency analysis. Several statistical regionalization methods proposed in the literature enable one to exploit sequences of precipitation extremes observed at homogeneous pooling groups of sites, that supposedly share the same frequency regime of rainfall extremes with the site of interest. Homogeneous sites can be identified by looking at specific climatic descriptors; for instance, some reliable authors successfully utilize Mean Annual Precipitation (MAP) as the sole proxy for locally characterizing the frequency regime of sub-daily rainfall extremes, and for grouping sequences of rainfall extremes records. We aim at advancing this traditional approach (1) by relaxing the hypothesis of the existence of a homogeneous pooling group of sites characterized by a unique regional parent distribution and (2) by incorporating additional morphological and climatic information in the regional model. We rely on more than 2350 Annual Maximum Series of rainfall depth for different time-aggregation intervals between 1 and 24 hours, observed since 1928 to 2011 in a vast study area in Northern Italy.  We refer to MAP as well as to additional morphologic descriptors (e.g.  minimum distance to Tyrrhenian (Adriatic) Sea, mean elevation and slope around the station, etc.).

We train a probabilistic neural network that models the frequency regime of observed annual maxima of rainfall depth resorting to a Generalized Extreme Value (GEV) distribution, whose parameters are data-driven functions of the local values of the selected descriptors and duration. Then, several cross-validation experiments are performed to assess the accuracy of the developed regional model relative to a simpler regional GEV model, whose parameters are functions of MAP and time-aggregation intervals.

Our analyses address several research problems: (a) identifying the most descriptive morphological proxies to combine with MAP for representing the frequency regime of sub-daily rainfall extremes in the study area, (b) highlighting limitations and potential of data-driven multivariate regional models of the frequency regime of rainfall extremes, (c) the advantages of a multivariate approach relative to a regionalization scheme based on MAP alone.

How to cite: Magnini, A., Lombardi, M., Valtancoli, E., and Castellarin, A.: Multivariate and data-driven regional frequency analysis for rainfall extremes, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-381, 2022.

Valentin Mansanarez, Benjamin Renard, and Michel Lang

This work presents trend analyses performed on time series. It follows the work done by Giuntoli et al. (2013) on trend analysis of low flows and their relationship with large-scale climate variability in France. A R package was implemented to allow the extraction of variables on time series and performing statistical trend analysis on the extracted variables. Detected trend can be summarised on maps. The methodology uses the Mann-Kendall statistical test to assess the significance of linear trend. Regional consistency can also be checked.

The methodology used was performed on 207 French daily streamflow over the period 1968-2020. Three period subsets are studied: 1968-2000, 1968-2010 and 1968-2020 to compare trend results and assess the variability over time.
Results confirm a North-South geographical split in temporal trends for droughts. They also show the increase of the severity of droughts over the last decades in Southern France.

Results also suggests a similar North-South geographical split for high and medium flows. They show that temporal trends are decreasing over the last decades in the Southern part of France for both high and medium flows. However, in the North part of France, results shows less significant trends in streamflow time series.

Giuntoli, I., Renard, B., Vidal J.-P., and Bard, A. (2013). Low flows in France and their relationship to large-scale climate indices, Journal of Hydrology, 482, 105-118, http://dx.doi.org/10.1016/j.jhydrol.2012.12.038.

How to cite: Mansanarez, V., Renard, B., and Lang, M.: A R package for quickly updating trend analysis: application to French streamflow time series.  , IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-647, 2022.

Ronald van Nooijen, Alla Kolechkina, and Changrang Zhou

A necessary first step in any hydrological study is the analysis of the available measurements and in particular their relevance to the future behaviour of the system. Periodic variations and linear trends are of special interest. However, an abrupt change in the system and the statistical processes that affect the measurements will, if uncorrected, interfere with the analysis. Detecting abrupt changes in time  series is therefore of considerable importance. The classical tests are primarily intended to test the null hypothesis “There is no change point”. If there is a change point (CP), then the usual approach reuses the classical test results to find a location. Additional work is then needed to determine the uncertainty in the position.

In 2018 two new approaches to change point analysis were presented in the statistical literature. One approach uses a deviance function and the other uses homogeneity tests. Both approaches construct confidence sets for the change point location at all confidence levels.  An interesting aspect of these approaches is that they can produce confidence sets, not just confidence intervals. For example,  in a series of annual maxima two points separated by several years may be marked as potential CP at a given confidence level without marking the intervening years as possible CPs at that level. The approach using homogeneity tests was applied to  both synthetic time series and measurement time series. The results for synthetic time series provide information on the statistical properties of the approach for small samples.  The results for measurement time series are compared with the CPs found for these series in the literature.

How to cite: van Nooijen, R., Kolechkina, A., and Zhou, C.: Building confidence curves with classical change point tests, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-714, 2022.

Arathy Nair, Sankaran Adarsh, G Mohan Meera, and Vi Sreedevi

Global climate changes significantly contribute to increased frequency of hydrologic extremes. This significantly underestimates the hydrologic design parameters, bringing of hydro systems to increased failure risk. In order to address this concern, the current practice of development of hydrologic frequency tools need to be updated accounting for non-stationarity. This study first considered a diverse set of statistical tests to examine the trend, change points, non-stationarity and randomness of streamflow, rainfall and temperature time series of scales ranging from daily to annual. The annual maxima time series indicted non stationarity against the stationary behaviour of daily series of hydro-meteorological datasets of the basin. Subsequently, this study developed the Temperature Duration Frequency (TDF), rainfall Intensity Duration Frequency (IDF) and flood frequency (FF) curves of Greater Pamba river basin in Kerala India, the part of which was most severely affected by the near century return period flood event of 2018. The analysis was performed for a multitude of combinations of variations in distribution parameters with time and climatic drivers as physical covariates in the extreme value formulations. The study proposed a novel wavelet coherence (WC) based driver selection of most dominant combination of climatic precursors in developing FF and IDF relations of three locations of Kalloopara, Malakkara and Thumpamon and TDF curve of Kuttanad region in the basin, considering data of 1978-2015 period. The proposed WC framework considers bi-multi-and partial effects of climatic oscillations (COs) like ElNino Southern Oscillation (ENSO), Indian Ocean Dipole (IOD), Pacific Decadal Oscillation (PDO) and North Atlantic Oscillation (NAO) in identifying potential drivers. The different WC formulations captured in-phase relationships of streamflows and rainfall with COs at intra-annual, annual and inter annual scales upto 4 years. The methods showed that addition of climatic precursors improved the NS estimates of flood and rainfall quantiles by more accurately capturing the magnitudes of extreme streamflows and rainfalls of 2018, 2021 than the time covariate formulations. However, the role of COs on extreme temperature is not found to be influential in developing TDF relationships, which needs further investigation.

How to cite: Nair, A., Adarsh, S., Meera, G. M., and Sreedevi, V.: Developing Non Stationary Frequency Relationships for Greater Pamba River basin, Kerala India incorporating dominant climatic precursors, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-743, 2022.

Coffee break
Chairperson: Alberto Viglione
Stelios Vavoulogiannis, Theano Iliopoulou, Panayiotis Dimitriadis, and Demetris Koutsoyiannis

Time asymmetry, i.e., temporal irreversibility, has a very important role in many scientific fields and has been studied thoroughly. Its detection in time series indicates the need to preserve it in stochastic simulations. This also seems to be the case for the streamflow process in hydrological simulation. Relevant large-scale studies have shown that time asymmetry of the streamflow is absolutely evident at small scales (hours) and vanishes only at larger scales (several days). The latter highlights the need to reproduce it in flood simulations of fine-scale resolution. To this aim, an enhancement of a recently proposed simulation algorithm for irreversible processes was developed, based on an asymmetric moving average (AMA) scheme that allows for the explicit preservation of time asymmetry at two or more timescales simultaneously. The method is tested through some case studies from around the world to further explore the method’s strengths and limitations and to examine the stochastic characteristics of the simulated results.

How to cite: Vavoulogiannis, S., Iliopoulou, T., Dimitriadis, P., and Koutsoyiannis, D.: Time Asymmetry and Stochastic Modelling of Streamflow, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-270, 2022.

Jorge Sebastian Moraga, Nadav Peleg, Peter Molnar, Simone Fatichi, and Paolo Burlando

Hydrological projections in the context of a changing climate may display high levels of uncertainty, particularly when examining small temporal and spatial scales. To project the response of hydrological processes to the increasing global temperatures, scientists and practitioners rely on chains of numerical models, each contributing some degree of uncertainty to the overall outputs. Furthermore, the randomness intrinsic to climate phenomena, known as internal climate variability, contributes to the uncertainty of the hydrological projections in the form of an irreducible stochasticity. In this work, we quantify the impacts and partition the uncertainty of hydrological processes emerging from climate models and internal variability  for two mountainous catchments in the Swiss Alps and across a broad range of scales. To that end, we used high-resolution ensembles of climate and hydrological data produced using a two-dimensional stochastic weather generator (AWE-GEN-2d) and a distributed hydrological model (Topkapi-ETH). We quantified the uncertainty in hydrological projections towards the end of the century through the estimation of the values of signal-to-noise ratios (STNR). We found small STNR values (<-1) in the projection of annual streamflow for most sub-catchments in both study sites that are dominated by the large natural variability of precipitation (explains ~70% of total uncertainty). Furthermore, we investigated specific hydrological components that are critical in the model chain with detail. For example, snowmelt or liquid precipitation exhibits robust change signals, which translates into high STNR values for streamflow during warm seasons and at higher elevations, together with a larger contribution of climate model uncertainty, suggesting that an improvement of the involved models has the potential of significantly narrowing the uncertainty. In contrast, extreme flows show low STNR values due to large internal climate variability across all elevations, which limits the possibility of narrowing their estimation uncertainty due to a warming climate. This study demonstrates that high-resolution hydro-climatic ensembles enable the quantification of hydrological projections across spatial and temporal scales, which can be used to assess the potential for narrowing hydrological uncertainties.

How to cite: Moraga, J. S., Peleg, N., Molnar, P., Fatichi, S., and Burlando, P.: Using high-resolution stochastic climate ensembles to model the impacts and uncertainty of hydrology in mountainous catchments, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-273, 2022.

Maria Elenius and Göran Lindström

Hydropower regulations may significantly increase the variability of flow when compared with the natural hydrological regime to which river ecosystems have evolved over long time periods. This can be detrimental for river habitats and for many organisms. Attenuation of the variability in rivers improves ecological status at some distance downstream of the introduced variability. Being able to accurately estimate this distance is critical for the evaluation of ecological status. The attenuation of introduced flow variability has only been studied previously for specific rivers, and the dominant mechanisms have not been analyzed in detail. In this work, the attenuation rate and its important drivers is studied for regulated rivers in all of Sweden by comparing the results of hydrological and hydrodynamic models with observations. We performed Fourier transformation of flow time series from the Hydrological Predictions for the Environment (HYPE) model, an extracted model representing only river processes, the diffusion wave equation, and from observed flow at several hundred stations. The reduction of the amplitudes along rivers was then analysed.

In many regulated rivers in Sweden, flow variability of periodicity 7 days is dominant among periods varying from a couple of days up to one month. The analysis further shows that variability with periodicity days to months typically attenuate with an exponential rate that is largest for short periods. Attenuation of these periods is mainly driven by processes within rivers, as opposed to catchment features such as the distribution of rain or soil properties. Further, rivers in regulated systems often resemble cascades with long stretches of rivers with low gradients in elevation between the dams. The associated attenuation in these “lake-alike” rivers can be well described by hydrological simulations with HYPE using a simple linear attenuation box. In contrast, the sometimes-used diffusion wave equation is often unable to replicate the observed attenuation here. Our work supports the assessment of ecological status and management decisions by improving the estimates of distances required for attenuation, and provides important insights on attenuation processes. Further, the analysis of dominant modes can be used to parameterize short-term regulations in hydrological simulations to improve their forecast skills.

How to cite: Elenius, M. and Lindström, G.: Dominant discharge periodicities and their attenuation in rivers of Sweden, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-358, 2022.

Roman Výleta, Peter Valent, and Ján Szolgay

When solving practical water resources management problems, situations when the amount of data is insufficient both terms of length and representativeness occur. Even the most extended observed series are short for reliable estimation and assessment long-term variability and extremes. A combination of stochastic weather generators, allowing the generation of arbitrarily long synthetic series of precipitation and air temperatures, and rainfall-runoff models, which transform them into an equally long synthetic series of river discharges, represent a feasible solution. This study aimed to develop a robust, practically oriented stochastic weather generator for rainfall-runoff modelling allowing the simulation of synthetic daily precipitation and air temperature series at multiple stations in catchments, considering seasonality, temporal dependency and spatial correlation. The design was based partly on the established methodological practices. In addition, it implements innovative components of temporal ad spatial multiscale disaggregation based process-oriented analysis of the processes involved and allows to consider the possibility of climate change. The ability of the weather generator to correctly reproduce characteristics of the observed rainfall and air temperature records was evaluated by both (temporal and spatial) statistical and hydrological process characteristics at each and across all stations.

Several case study results proved that the proposed concept of the generator is robust and practically applicable. It allows to reliably generate synthetic series of precipitation totals and air temperatures at individual stations in catchments and around simultaneously. Using a combined stochastic-deterministic rainfall-runoff modelling approach provides an infinite number of combinations of flow, including the extreme and the unobserved ones. It can be used to estimate extreme flood characteristics, determine hydrologic metrics of ecological flows, and detect changes in flow variability caused by land use and climate change.

Acknowledgements: This work was supported by the Slovak Research and Development Agency under Contract No. APVV-19-0340 and the VEGA Grant Agency No. 1/0632/19.

How to cite: Výleta, R., Valent, P., and Szolgay, J.: A process-based multisite multivariate stochastic weather generator for water resources management, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-362, 2022.

Demetris Koutsoyiannis and Alberto Montanari

We present a new method for simulating and predicting hydrologic variables with uncertainty assessment and provide example applications to river flows. The method is identified with the acronym ``Bluecat'' and is based on the use of a deterministic model which is subsequently converted to a stochastic formulation. The latter provides an adjustment on statistical basis of the deterministic prediction along with its confidence limits. The distinguishing features of the proposed approach are the ability to infer the probability distribution of the prediction without requiring strong hypotheses on the statistical characterization of the prediction error (e.g. normality, homoscedasticity) and its transparent and intuitive use of the observations. Bluecat makes use of a rigorous theory to estimate the probability distribution of the predictand conditioned by the deterministic model output, by inferring the conditional statistics of observations. Therefore Bluecat bridges the gaps between deterministic (possibly physically-based, or deep learning-based) and stochastic models as well as between rigorous theory and transparent use of data with an innovative and user oriented approach. We present two examples of application to the case studies of the Arno river at Subbiano and Sieve river at Fornacina. The results confirm the distinguishing features of the method along with its technical soundness. We provide an open software working in the R environment, along with help facilities and detailed instructions to reproduce the case studies presented here.

How to cite: Koutsoyiannis, D. and Montanari, A.: Bluecat: A Local Uncertainty Estimator for Deterministic Simulations and Predictions, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-574, 2022.

Auguste Gires, Jerry Jose, Daniel Schertzer, and Ioulia Tchiguirinskaia

Understanding rainfall processes remains a major challenge to hydrology. As an illustration, it should be reminded that the lack of precise knowledge on rainfall is one of the greatest sources of uncertainty in hydrological modelling. In the context of this presentation, rainfall should not be viewed as a simple rainfall rate but as a complete drop size distribution (DSD). Since rain is obviously an atmospheric field, a more in-depth understanding of it requires studying its relationship to wind turbulence. It is well known that rainfall and wind turbulence exhibit extreme variability over wide ranges of spatio-temporal scales. Hence, a framework for managing such characteristics must be implemented to study their relationships. Universal Multifractals that are parsimonious, physically based and mathematically robust are a suitable framework. In this presentation, we suggest to investigate correlations across scales between rainfall and wind, depending on various meteorological conditions. One key question is whether or not raindrops behave like passive scalars.

To do this, the authors will benefit from atmospheric data collected during a campaign of high resolution measurement on a meteorological mast. More precisely, data collected in the framework of the RW-Turb project (supported by the French National Research Agency – ANR-19-CE05-0022) by 3D sonic anemometers (manufactured by Thies), mini meteorological stations (manufactured by Thies), and disdrometers (Parsivel2, manufactured by OTT) installed at two heights (approx. 45 m and 80 m), which enables to monitor potential effects of altitude will be used. The temporal resolution is of 100 Hz for the 3D sonic anemometers, 1 Hz for the meteorological stations and 30 s for the disdrometers.

Initial developments of a stochastic 3+1D model for the droplet field based on the previous results will be presented as well as its consequences on rainfall measurements notably with weather radar which are the only device measuring rainfall in both space and time. Finally, the consequences of the correlations recovered through the scales on hydrology and wind energy production will be discussed.

How to cite: Gires, A., Jose, J., Schertzer, D., and Tchiguirinskaia, I.: Universal Multifractal relations between rainfall and wind turbulence, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-663, 2022.

Lunch Break // W4F Hackathon High Schools Final
Chairperson: Elena Volpi
Antonio Zarlenga, Mariaines Di Dato, Claudia D'Angelo, and Alessandro Casasso

Shallow geothermal systems represent a unique opportunity for heating and cooling of buildings with green energy and low operational costs. So far, the geothermal system studies have been built on simplifying assumption such as homogeneous porous medium or purely advective flow. However, solutions based on equivalent conductivity and effective macrodispersion coefficients may not be able to grasp the effects of aquifer heterogeneity on geothermal systems.

The aim of our research is to investigate how thermo-hydrological and engineering parameters impact the different heat transport dynamics and how they result in the GS efficiency. The study considers an open loop GS made by a well doublet placed into a confined heterogeneous aquifer of constant thickness; groundwater is abstracted upstream and, after the heat exchange, it is reinjected downstream with a constant altered temperature. The efficiency of GS is intrinsically related to the behavior of the breakthrough time distribution, i.e. the distribution of the travel time the water particles employ to move from the injection to the extraction well, and the recirculating ratio. By means of accurate numerical simulations that mimic the heat transport through a Lagrangian procedure we identify and explore the effect of hydraulic conductivity heterogeneity, pore scale dispersion, thermal diffusion, pumping rate and geometrical parameters.

The analysis hints that the hydraulic conductivity heterogeneity has a strong impact on the early operational time: due to channeling, the thermal plume travels faster in the highly conductive layers. As a result, the first breakthrough time, the key parameter adopted in the design of GS, decreases with heterogeneity, moreover, the uncertainty associated with early arrivals increases with heterogeneity.

The heterogeneity, as well as dispersion and convection, has a negligible effect on the long-term period.The recirculating ratio depends strongly on the pumping rate and other geometrical parameters.

Given that well screens usually cross a short depth we perform a detailed analysis on the uncertainty related to the ergodicity issue. Result of a single realization can significantly differ from its ergodic counterpart. As a practical consequence, a thermal feedback occurring in a heterogeneous medium could significantly differ from the expected theoretical one.

How to cite: Zarlenga, A., Di Dato, M., D'Angelo, C., and Casasso, A.: Efficiency of shallow geothermal systems in heterogeneous media, how to manage uncertainty, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-357, 2022.

Charles Onyutha

Hydrological models are commonly applied to investigate impacts of climate change and variability on hydrology. Confidence in hydrological predictions is linked to the model’s performance in reproducing observed variable under consideration. Judgment of a model’s quality is challenged by the differences which exist among the various efficiency criteria. Some of the efficiency criteria commonly applied for assessment of hydrological models include Nash-Sutcliffe Efficiency, coefficient of determination and Kling and Gupta Efficiency. However, a plethora of recently introduced "goodness-of-fit" metrics exists including, for instance, the revised R-squared and Liu efficiency. Studies on comparative analysis of the various efficiency criteria applied to hydrological modeling considering both the old and recently introduced "goodness-of-fit" metrics are lacking. In this study, several efficiency criteria are compared with respect to the computation difficulty, and speed of computation while considering various sample sizes. The pros and cons of the various (both old and new) efficiency criteria are given. The results of this study show the need for modelers to gain insights into the impact of each "goodness-of-fit" metric on model performance assessment before making hydrological predictions for planning predictive adaptations to the impacts of climate change or variability on hydrology.

How to cite: Onyutha, C.: Pros and cons of various efficiency criteria for evaluating hydrological models, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-485, 2022.

Pau Wiersma, Fatemeh Zakeri, and Grégoire Mariéthoz

To study long-term changes in the hydrology of snow-fed catchments, there is a need for long-term time series of snow cover data. Satellite imagery and climate reanalysis can both be used to quantify past snow cover, but they lack in baseline period length and spatial resolution respectively. In this study we apply statistical methods to generate synthetic high-resolution daily snow cover maps, and consequently use these maps as forcing in long-term hydrological modeling. The results are benchmarked against a case without synthetic snow cover forcing. Multiple hydrological models are used to reduce the uncertainty related to the model choice. The study is performed on the Thur catchment in Eastern Switzerland, a meso-scale catchment covering a wide elevation range and experiencing multiple periods of intermittent snow cover annually. We expect the synthetic snow cover maps to provide added value through the high-resolution spatial information on snow appearance and disappearance, leading to better estimates of snow melt runoff. However, we also expect them to show some physical inconsistencies, particularly after periods of high snow accumulation. Should it prove promising, this approach can be used to study both past and future hydrological changes in any snow-fed catchment using ERA5 and CMIP6 climate data, potentially spanning the entire period 1950-2100. Additionally, this framework of synthetic satellite data generation could be expanded to other hydrological variables or to environmental image time series in other fields than hydrology.

How to cite: Wiersma, P., Zakeri, F., and Mariéthoz, G.: The value of synthetic high-resolution daily snow cover maps for long-term hydrological modeling, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-535, 2022.

Rodric Merimé Nonki, Ernest Amoussou, Raphael Muamba Tshimanga, Houteta Djan'na Koubodana, Franck Eitel Kemgang Ghomsi, and André Lenouo

Conceptual rainfall-runoff models are commonly used in many hydrological applications in support of water resources management practices. They provide an advantage in data scarce areas due to their ability to use limited data and generate sufficiently reliable information. The main challenge with this type of hydrological models remains the ability to establish an optimal model parameter space due to a number of sources of uncertainties. This study is conducted in the Upper Benue River Basin (UBRB) in Cameroon with the aim to establish a rainfall-runoff model that is fit in the context of hydro-climate characteristics of the basin. The UBRB is the second-largest river in Cameroon and one of the most important water resources in northern Cameroon from both a water supply and hydropower generation perspective. A Monte Carlo procedure was implemented to calibrate the HYMOD conceptual rainfall-runoff model using five statistical performance measures: the Absolute percent bias (PBIAS), the ratio between root mean square error (RMSE) and standard deviation of observed data (STDEVobs) (RSR), the Pearson correlation coefficient (r), the Nash–Sutcliffe efficiency (NSE) and the Kling–Gupta efficiency (KGE). The results reveal that the model performance varies from good to very good with NSE greater than 0.78 and RSR less than 0.46 during the calibration and validation periods. Therefore, this model can be used to support various water resources management initiatives in the basin.

How to cite: Nonki, R. M., Amoussou, E., Tshimanga, R. M., Koubodana, H. D., Kemgang Ghomsi, F. E., and Lenouo, A.: Performance assessment of a daily time-step HYMOD conceptual rainfall-runoff model for the Upper Benue River, Cameroon, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-547, 2022.

Tejith Pogakula and Tirupati Bolisetti

The climate model projections obtained from the Regional Climate Models (RCMs) and their forcing through the hydrological models are known to include multi-source uncertainties. Using the subsequent modelling data for their intended purposes, such as watershed studies, flood mitigation, and climate change adaptation policies while containing unsupervised uncertainties of such nature, will prove detrimentally misjudged and potentially do more harm than good. The uncertainty propagation takes place at each stage along the modelling process and depends on the choice of climate model projections, Representative Concentration Pathways (RCPs), bias correction methods (BCs),  hydrological models, and hydrological parameters among other contributors.


The aim of this paper is to quantify the overall uncertainty in climate change impacts using Analysis of Variance (ANOVA) to decompose or disaggregate it into components and assess relative contribution of each of the components. The approach is demonstrated through a case study in Little River Experimental Watershed (LREW watershed) under two different emission scenarios (RCP 4.5 and RCP 8.5), five sets of RCM and driving GCM combinations, two bias correction methods by utilizing the Soil & Water Assessment Tool (SWAT) hydrological model. The results will indicate breakdown of the total uncertainty (T) into the respective uncertainties caused by the climate models (C), emission scenario (R), bias correction method (B) and unlike most of the uncertainty decomposition studies, further uncertainty breakdown due to the interactions between various components are also presented. The findings from this study will be useful to the modelers involved with flood mitigation or policy management by enhancing their understanding about the nature of streamflow projections and effectively aid in better decisions concerned with adapting to a changing climate subjected to uncertainties.

How to cite: Pogakula, T. and Bolisetti, T.: Uncertainty quantification in climate change impacts on hydrology using ANOVA, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-571, 2022.

Kolbjørn Engeland, Thea Roksvåg, Stein Beldring, Erik Holmqvist, Anja Iselin Pedersen, and Gusong Ruan

We present a new runoff map for Norway for the reference period 1991-2020. A new framework combing precipitation runoff-modelling with geostatistical interpolation was used. The precipitation-runoff model WASMOD was used to simulate runoff on a 1x1 km grid covering all Norway and nearby catchments in Sweden and Finland. The parameters of WASMOD were conditioned on land-use and climate classes and calibrated globally using data from 215 streamflow stations. Figure 1a) shows the runoff estimates from WASMOD that are biased when compared to observations (Figure 2a) .

To correct the biases in the gridded simulations we used runoff observations from 732 locations of which 198 had data covering the whole period. A geostatistical approach was used to estimate the 30 year mean annual runoff from short records (1-29 years) for 482 catchments. For 45 catchments with substantial glacier coverage or regulation capacity, a manual extension of mean annual runoff was performed. Another 7 stations with between 25 and 30 years of data were included.

The biases from WASMOD were corrected by simple linear regression using the raw runoff map as a covariate and the runoff observations as the dependent variable. The regression coefficients were modelled as spatial fields using a geostatistical Bayesian approach where SPDE and INLA ensured fast Bayesian inference. The mean annual runoff after the correction is shown in Figure 1b), and in Figure 1c) the difference between the two previous maps is shown. Figure 2b shows a scatter plot of corrected and observed runoff. The scatter is now much smaller.

Figure 1. Mean annual runoff from a) the Wasmod model and b) after the correction. The difference in runoff between a) and b) is shown in c)

The new framework was evaluated by cross-validation. On average this new approach outperformed a WASMOD when predicting runoff for ungauged and partially gauged catchments and reduced the RMSE by 42% . Scatter plot of predicted runoff is shown in Figure 2c).

Figure 2. Scatterplot of mean annual runoff for a) WASMOD, b) corrected values, and c) predicted runoff , versus observations in 146 catchments in a split sample cross-validation test




How to cite: Engeland, K., Roksvåg, T., Beldring, S., Holmqvist, E., Pedersen, A. I., and Ruan, G.: New runoffmap for Norway by combining a precipitation runoff model with geostatistical interpolation, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-723, 2022.

Posters: Wed, 1 Jun, 15:00–16:30 | Poster area

Chairpersons: Antonio Zarlenga, Elena Volpi
Nejc Bezak, Matjaž Mikoš, Klaudija Lebar, Simon Rusjan, and Mojca Šraj

Engineers are often required to derive design peak discharges and hydrographs that are needed for the planning, design and construction of the hydro-technical structures such as dikes, dams or preparation of the flood risk or hazard maps. There are numerous options available for the design peak discharge and hydrograph estimation ranging from simple empirical equations to complex rainfall-runoff models including stochastic rainfall simulators. Additionally, rainfall-runoff model parameters can be estimated using a variety of methods depending on the availability and quality of data. However, such methods are often too complex or time consuming to be used in practice. Therefore, the knowledge gap between theoretical hydrology and practical hydrological applications can be very large in some cases. Thus, engineers are often using very simple methods that do not account for the uncertainty in the design values estimation. Additionally, they often do not use all available data (e.g., discharge, precipitation).

Moreover, a step forward is also needed in the process of the preparation of the flood hazard and risk maps that have been prepared across Europe based on requirements of the EU Flood Directive (Directive 2007/60/60). In order to follow good practices in some other fields (e.g., earthquake engineering), where probabilistic approach is used in the design, a similar methodology would need to be developed for the preparation of flood maps taking into account stochastic nature of the hydrological processes.

Therefore, a robust methodology that takes into account the latest state-of-the-art research findings and knowledge is needed to guide engineers in the process of the design discharge and hydrograph definition. Such methodology is currently lacking in Slovenia, where complexity of hydrological estimations is often on lower level compared to the hydraulic calculations conducted. This contribution presents recent work carried out in relation to the definition of the methodology for the design peak discharge and hydrograph calculation in case of gauged and ungauged catchments in Slovenia. The methodology will also include guidelines related to the stochastic simulations and definition of multiple design hydrographs that could be used to derive probabilistic flood risk maps.

We acknowledge the financial support from the Slovenian Research Agency (grant: V2-2137).

How to cite: Bezak, N., Mikoš, M., Lebar, K., Rusjan, S., and Šraj, M.: Towards development of the methodology for the design hydrograph definition including stochastic simulations in Slovenia, Europe, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-13, 2022.

Silvia Salas Aguilar, Etienne Leblois, Jean-Dominique Creutin, and Enrique Gonzalez Sosa

Many stochastic rainfall generators rely on the structure of rainfall considered a spatio-temporal point process. Ordinary practice is to infer these parameters from data observed on a group of rain gauges over some reference time. However, gridded estimates of rainfall (out of meteorological radar, satellites, or meteorological models) give a distinct perspective on the same precipitation, albeit with scaling issues and biases, and a question is how to value this information even if we target the urban scale. This contribution is about how the scaling issue can be handled in a geostatistical perspective, deriving a point variability model consistent with the variability observed at grid scale. Available local raingauge data, even in moderate amount, can be used to checked the suggested methodology. Expected uses are in data merging, downscaling climate scenarios, and establishing local rainfall models, also in local data-poor contextes.



How to cite: Salas Aguilar, S., Leblois, E., Creutin, J.-D., and Gonzalez Sosa, E.: Estimating point rainfall statistics from gridded large scale rainfall data, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-168, 2022.

Faranak Tootoonchi, Mojtaba Sadegh, Jan Olaf Haerter, Olle Räty, Thomas Grabs, and Claudia Teutschbein

A warming climate is associated with increasing hydroclimatic extremes, which are often interconnected through complex processes, prompting their concurrence and/or succession, and causing compound extreme events. It is critical to analyze the risks of compound events, given their disproportionately high adverse impacts. To account for the variability in two or more hydroclimatic variables (e.g., temperature and precipitation) and their dependence, a rising number of publications focuses on multivariate analysis, among which the notion of copula-based probability distribution has attracted tremendous interest. Copula is a mathematical function that expresses the joint cumulative probability distribution of multiple variables. Our focus is to re-emphasize the fundamental requirements and limitations of applying copulas. Confusion about these requirements may lead to misconceptions and pitfalls, which can potentially compromise the robustness of risk analyses for environmental processes and natural hazards. We conducted a systematic literature review of copulas, as a prominent tool in the arsenal of multivariate methods used for compound event analysis, and underpinned them with a hydroclimatic case study in Sweden to illustrate a practical approach to copula-based modeling. Here, we (1) provide end-users with a didactic overview of necessary requirements, statistical assumptions and consequential limitations of copulas, (2) synthesize common perceptions and practices, and (3) offer a user-friendly decision support framework to employ copulas, thereby support researchers and practitioners in addressing hydroclimatic hazards, hence demystify what can be an area of confusion.

How to cite: Tootoonchi, F., Sadegh, M., Haerter, J. O., Räty, O., Grabs, T., and Teutschbein, C.: Copulas for hydroclimatic analysis: A practice-oriented overview, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-343, 2022.

Prediction of daily water temperature in the Yangtze River based on the C-vine copula
Yuwei Tao, Yuankun Wang, and Dong Wang
Dario Treppiedi, Gianluca Sottile, Giada Adelfio, and Leonardo Valerio Noto

Physical modelling of atmospheric processes, such as rainfall events, is often difficult due to the complexity of the atmosphere and the large number of variables involved. At the same time, it is necessary to know, as carefully as possible, some characteristics of precipitation processes, such as rainfall magnitude and frequency, in order to better understand their impacts on the territory. For this reason, statistical frameworks able to use external covariates to explain the physical process could be a central point in research.

Starting from the rainfall events identified from continuous data series from about 40 rain gauges located in Sicily, this paper aims at assessing the occurrence of rainfall events in a fixed interval of time according to a temporal contagion model (branching process) with external covariates, within a regression-like framework (Adelfio and Chiodi, 2021).

In detail, we extend the model formulation proposed by Meyer et al. (2012) in the context of infectious disease transmission, suggesting the use of a specific branching-type model, born in seismic context (the ETAS model, (Ogata 1988, 1998)), in a regression-oriented version modelling. In the temporal ETAS model, the expected frequency of events in a time unit can be defined as the sum of a term that describes the long-term variation and a term that describes the short-term variation.

Accounting for further potential covariates in the model specification of the short-term variation component, may both explain some of the overall variability of the studied phenomenon (i.e., for decreasing the unpredictable variability) and provide a more realistic description of the observed activity. The Forward Likelihood for prediction (FLP) method (Chiodi and Adelfio 2011) is used for estimating the ETAS model components with the covariates.

In this application the mean rainfall intensity, the duration and the anomalies in temperature and relative humidity of the events have been considered as external covariates of the model in order to explain the events frequency. The first results of the model appear to be interesting, and special attention will be paid to the sample of convective precipitation events identified using the same dataset (Sottile et al. 2021).

How to cite: Treppiedi, D., Sottile, G., Adelfio, G., and Noto, L. V.: Modelling rainfall event frequency through a temporal contagion model with external covariates, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-413, 2022.

Salvatore Grimaldi, Elena Volpi, Andreas Langousis, Simon Michael Papalexiou, Davide Luciano De Luca, Rodolfo Piscopia, Sofia D. Nerantzaki, Georgia Papacharalampous, and Andrea Petroselli

The benefit of continuous modelling in hydrological studies is widely recognized since it provides the practitioner with effective hydrological outputs for risk assessment. However, this approach is still not common mainly because it needs as input simulated rainfall time series. Many rainfall generation methods exist yet there are still two main challenges: (i) many rainfall models are available without a clear suggestion on which is the most appropriate to use, (ii) most rainfall models are not user-friendly and require significant theoretical background to successfully be applied.

In this contribution, we test eight rainfall models by evaluating the performances of the simulated rainfall time series when used as input for a specific continuous rainfall-runoff model, named COSMO4SUB (COntinuous Simulation MOdel For Small and Ungauged Basin), particularly designed for small and ungauged basins. The rainfall models selected here are: two versions of the Complete Stochastic Modelling Solution (CoSMoS-1s and 2s); three versions of Bootstrap-based models; the classical Bartlett Lewis and Neymann-Scott rectangular pulses models; and a mixed method based on monthly simulation and multifractal cascade disaggregation.

The comparison was performed by analyzing runoff time series obtained with the COSMO4SUB model and using as input the rainfall time series simulated by the eight models and the observed one. We selected and investigated several general properties, such as the average number of flood events, the marginal distribution of peak, volume, duration and antecedent dry period before the flood, and their dependence structure.

The comparison confirms the capability of all models to provide realistic flood events and allows identifying the models to be further improved and tailored for data scarce hydrological risk applications.

How to cite: Grimaldi, S., Volpi, E., Langousis, A., Papalexiou, S. M., De Luca, D. L., Piscopia, R., Nerantzaki, S. D., Papacharalampous, G., and Petroselli, A.: Rainfall model comparison for continuous modelling for small and ungauged basins, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-451, 2022.

Stefano Farris, Roberto Deidda, Francesco Viola, and Giuseppe Mascaro

According to both theoretical considerations and climatic projections, precipitation extremes are expected to increase under a warmer environment. Inferential analyses involving statistical testing procedures are frequently performed to validate this scenario. Recent research has found that the results of trend tests applied to hydrological data might be misinterpreted if (i) records exhibit autocorrelation and (ii) field significance is not taken into account when tests are performed multiple times. In this study, we investigate these two issues focusing on frequencies (or counts) of daily rainfall extremes. To this end, a sample of extreme precipitation frequency time series is derived from long-term (100-year) daily precipitation records retrieved by 1087 rain gauges belonging to the Global Historical Climate Network database. Several Monte Carlo simulations are performed involving random synthetic frequency time series generated through the Poisson first-order Integer-valued AutoRegressive model (Poisson-INAR(1)), reproducing the statistical properties of the observed counts and characterized by different sample size, autocorrelation level, and trend magnitude. The following are the key findings. (1) While empirical autocorrelations are likely due to the existence of trends, empirical trends cannot be explained solely by autocorrelation, suggesting that accounting for serial correlation may have a limited influence on trend analyses of extreme frequency time series. (2) Taking field significance into account enhances the interpretation of test results by reducing the type-I errors. (3) Parametric statistical trend tests based on linear and Poisson regression prove more powerful than non-parametric tests (e.g. Mann-Kendall test) in analyzing count series. (4) Finally, we use these insights to conduct trend assessments on observed counts, finding several clear spatial patterns of statistically significant increasing (decreasing) trends, mostly located in central and eastern United States and Northern Eurasia (southwestern United States, southern Europe, southern parts of Australia).

How to cite: Farris, S., Deidda, R., Viola, F., and Mascaro, G.: How much do serial correlation and field significance affect trend detection on extreme precipitation frequencies?, IAHS-AISH Scientific Assembly 2022, Montpellier, France, 29 May–3 Jun 2022, IAHS2022-636, 2022.