Günter Blöschl, Jan Szolgay, Juraj Parajka, Silvia Kohnová and Pavol Miklánek
Thomas Nester, Jürgen Komma and Günter Blöschl
This paper reports on experience with developing the flood forecasting model for the Upper Danube basin and its operational use since 2006. The model system consists of hydrological and hydrodynamic components, and involves precipitation forecasts. The model parameters were estimated based on the dominant processes concept. Runoff data are assimilated in real time to update modelled soil moisture. An analysis of the model performance indicates 88% of the snow cover in the basin to be modelled correctly on more than 80% of the days. Runoff forecasting errors decrease with catchment area and increase with forecast lead time. The forecast ensemble spread is shown to be a meaningful indicator of the forecast uncertainty. During the 2013 flood, there was a tendency for the precipitation forecasts to underestimate event precipitation and for the runoff model to overestimate runoff generation which resulted in, overall, rather accurate runoff forecasts. It is suggested that the human forecaster plays an essential role in interpreting the model results and, if needed, adjusting them before issuing the forecasts to the general public.
Elena Szolgayová, Josef Arlt, Günter Blöschl and Ján Szolgay
Short term streamflow forecasting is important for operational control and risk management in hydrology. Despite a wide range of models available, the impact of long range dependence is often neglected when considering short term forecasting. In this paper, the forecasting performance of a new model combining a long range dependent autoregressive fractionally integrated moving average (ARFIMA) model with a wavelet transform used as a method of deseasonalization is examined. It is analysed, whether applying wavelets in order to model the seasonal component in a hydrological time series, is an alternative to moving average deseasonalization in combination with an ARFIMA model. The one-to-ten-steps-ahead forecasting performance of this model is compared with two other models, an ARFIMA model with moving average deseasonalization, and a multiresolution wavelet based model. All models are applied to a time series of mean daily discharge exhibiting long range dependence. For one and two day forecasting horizons, the combined wavelet - ARFIMA approach shows a similar performance as the other models tested. However, for longer forecasting horizons, the wavelet deseasonalization - ARFIMA combination outperforms the other two models. The results show that the wavelets provide an attractive alternative to the moving average deseasonalization.
Daniel Skublics, Günter Blöschl and Peter Rutschmann
The Bavarian Danube River has experienced numerous large flood events in recent years which make flood management an urgent matter. The propagation of flood waves along the river is heavily influenced by controlled and natural flood retention. Over the past centuries, natural flood retention areas were lost due to river training, and the hydraulic characteristics of the channel-flood plain system were modified. The purpose of this paper is to understand the effect of river training on the flood retention characteristics along the Bavarian Danube. Systematic two-dimensional hydrodynamic modelling shows that extreme floods are attenuated more strongly in the present state of the channel-flood plain system than they were historically. This is because the retention areas are filled later during the event, so the attenuation effect is much larger for the same magnitude of the retention volume. Natural flood retention is therefore not an effective management option for reducing extreme floods on the Bavarian Danube. Controlled flood retention measures provide a higher efficiency regarding peak attenuation to retention volume ratio. On the other hand, the delay of flood peaks due to natural retention may be beneficial for the superposition of the flood waves with contributions from downstream tributaries.
Alberto Viglione, Magdalena Rogger, Herbert Pirkl, Juraj Parajka and Günter Blöschl
Since the beginning of hydrological research hydrologists have developed models that reflect their perception about how the catchments work and make use of the available information in the most efficient way. In this paper we develop hydrologic models based on field-mapped runoff generation mechanisms as identified by a geologist. For four different catchments in Austria, we identify four different lumped model structures and constrain their parameters based on the field-mapped information. In order to understand the usefulness of geologic information, we test their capability to predict river discharge in different cases: (i) without calibration and (ii) using the standard split-sample calibration/ validation procedure. All models are compared against each other. Results show that, when no calibration is involved, using the right model structure for the catchment of interest is valuable. A-priori information on model parameters does not always improve the results but allows for more realistic model parameters. When all parameters are calibrated to the discharge data, the different model structures do not matter, i.e., the differences can largely be compensated by the choice of parameters. When parameters are constrained based on field-mapped runoff generation mechanisms, the results are not better but more consistent between different calibration periods. Models selected by runoff generation mechanisms are expected to be more robust and more suitable for extrapolation to conditions outside the calibration range than models that are purely based on parameter calibration to runoff data.
Silvia Kohnová, Ladislav Gaál, Tomáš Bacigál, Ján Szolgay, Kamila Hlavčová, Peter Valent, Juraj Parajka and Günter Blöschl
The case study aims at selecting optimal bivariate copula models of the relationships between flood peaks and flood volumes from a regional perspective with a particular focus on flood generation processes. Besides the traditional approach that deals with the annual maxima of flood events, the current analysis also includes all independent flood events. The target region is located in the northwest of Austria; it consists of 69 small and mid-sized catchments. On the basis of the hourly runoff data from the period 1976- 2007, independent flood events were identified and assigned to one of the following three types of flood categories: synoptic floods, flash floods and snowmelt floods. Flood events in the given catchment are considered independent when they originate from different synoptic situations. Nine commonly-used copula types were fitted to the flood peak - flood volume pairs at each site. In this step, two databases were used: i) a process-based selection of all the independent flood events (three data samples at each catchment) and ii) the annual maxima of the flood peaks and the respective flood volumes regardless of the flood processes (one data sample per catchment). The goodness-of-fit of the nine copula types was examined on a regional basis throughout all the catchments. It was concluded that (1) the copula models for the flood processes are discernible locally; (2) the Clayton copula provides an unacceptable performance for all three processes as well as in the case of the annual maxima; (3) the rejection of the other copula types depends on the flood type and the sample size; (4) there are differences in the copulas with the best fits: for synoptic and flash floods, the best performance is associated with the extreme value copulas; for snowmelt floods, the Frank copula fits the best; while in the case of the annual maxima, no firm conclusion could be made due to the number of copulas with similarly acceptable overall performances. The general conclusion from this case study is that treating flood processes separately is beneficial; however, the usually available sample size in such real life studies is not sufficient to give generally valid recommendations for engineering design tasks.
Maria M. Cárdenas Gaudry, Dieter Gutknecht, Juraj Parajka, Rui A.P. Perdigão and Günter Blöschl
The aim of this study is to understand the seasonalities of runoff and precipitation and their controls along two transects in Peru and one transect in Austria. The analysis is based on daily precipitation data at 111 and 61 stations in Peru and Austria, respectively, and daily discharge data at 51 and 110 stations. The maximum Pardé coefficient is used to quantify the strength of the seasonalities of monthly precipitation and runoff. Circular statistics are used to quantify the seasonalities of annual maximum daily precipitation and annual maximum daily runoff. The results suggest that much larger spatial variation in seasonality in Peru is because of the large diversity in climate and topography. In the dry Peruvian lowlands of the North, the strength of the monthly runoff seasonality is smaller than that of precipitation due to a relatively short rainy period from January to March, catchment storage and the effect of upstream runoff contributions that are more uniform within the year. In the Peruvian highlands in the South, the strength of the monthly runoff seasonality is greater than that of precipitation, or similar, due to relatively little annual precipitation and rather uniform evaporation within the year. In the Austrian transect, the strength of the runoff seasonality is greater than that of precipitation due to the influence of snowmelt in April to June. The strength of monthly regime of precipitation and runoff controls the concentration of floods and extreme precipitation in Peruvian transects. The regions with strong monthly seasonality of runoff have also extreme events concentrated along the same time of the year and the occurrence of floods is mainly controlled by the seasonality of precipitation. In Austria, the monthly runoff maxima and floods occur in the same season in the Alps. In the lowlands, the flood seasonality is controlled mainly by summer extreme precipitation and its interplay with larger soil moisture. The analyses of precipitation and runoff data along topographic gradients in Peru and Austria showed that, overall, in Peru the spatial variation in seasonality is much larger than in Austria. This is because of the larger diversity in climate and topography.
Ján Szolgay, Ladislav Gaál, Tomáš Bacigál, Silvia Kohnová, Kamila Hlavčová, Roman Výleta, Juraj Parajka and Günter Blöschl
This paper analyses the bivariate relationship between flood peaks and corresponding flood event volumes modelled by empirical and theoretical copulas in a regional context, with a focus on flood generation processes in general, the regional differentiation of these and the effect of the sample size on reliable discrimination among models. A total of 72 catchments in North-West of Austria are analysed for the period 1976–2007. From the hourly runoff data set, 25 697 flood events were isolated and assigned to one of three flood process types: synoptic floods (including long- and short-rain floods), flash floods or snowmelt floods (both rain-on-snow and snowmelt floods). The first step of the analysis examines whether the empirical peak-volume copulas of different flood process types are regionally statistically distinguishable, separately for each catchment and the role of the sample size on the strength of the statements. The results indicate that the empirical copulas of flash floods tend to be different from those of the synoptic and snowmelt floods. The second step examines how similar are the empirical flood peak-volume copulas between catchments for a given flood type across the region. Empirical copulas of synoptic floods are the least similar between the catchments, however with the decrease of the sample size the difference between the performances of the process types becomes small. The third step examines the goodness-of-fit of different commonly used copula types to the data samples that represent the annual maxima of flood peaks and the respective volumes both regardless of flood generating processes (the traditional engineering approach) and also considering the three process-based classes. Extreme value copulas (Galambos, Gumbel and Hüsler-Reiss) show the best performance both for synoptic and flash floods, while the Frank copula shows the best performance for snowmelt floods. It is concluded that there is merit in treating flood types separately when analysing and estimating flood peak-volume dependence copulas; however, even the enlarged dataset gained by the process-based analysis in this study does not give sufficient information for a reliable model choice for multivariate statistical analysis of flood peaks and volumes.
Ladislav Gaál, Ján Szolgay, Tomáš Bacigál, Silvia Kohnová, Kamila Hlavčová, Roman Výleta, Juraj Parajka and Günter Blöschl
This paper analyses the bivariate relationship between flood peaks and corresponding flood event volumes modelled by empirical copulas in a regional context in the North-West of Austria. Flood data of a total of 69 catchments in the region are analysed for the period 1976–2007. In order to increase the sample size and the homogeneity of the samples for the statistical analysis, 24872 hydrologically independent flood events were isolated and assigned to one of three flood process types: synoptic floods, flash floods or snowmelt floods in contrary to the more traditional engineering approach of selecting annual maxima of flood peaks and corresponding flood volumes. The first major part of the paper examines whether the empirical peak-volume copulas of different flood process types are statistically distinguishable, separately for each catchment. The results indicate that the empirical copulas of flash floods tend to be different from those of the synoptic and snowmelt floods in the target region. The second part examines how similar are the empirical flood peak-volume copulas between catchments for a given flood type. For the majority of catchment pairs, the empirical copulas of all flood types are indeed statistically similar. The flash floods show the largest degree of spatial heterogeneity. It is concluded that there is merit in treating flood types separately and in pooling events of the same type in a region when analysing and estimating flood peak-volume dependence copulas; however, the sample size of the analysed events is a limiting factor in spite of the introduced event selection procedure.
Mojca Šraj, Alberto Viglione, Juraj Parajka and Günter Blöschl
Substantial evidence shows that the frequency of hydrological extremes has been changing and is likely to continue to change in the near future. Non-stationary models for flood frequency analyses are one method of accounting for these changes in estimating design values. The objective of the present study is to compare four models in terms of goodness of fit, their uncertainties, the parameter estimation methods and the implications for estimating flood quantiles. Stationary and non-stationary models using the GEV distribution were considered, with parameters dependent on time and on annual precipitation. Furthermore, in order to study the influence of the parameter estimation approach on the results, the maximum likelihood (MLE) and Bayesian Monte Carlo Markov chain (MCMC) methods were compared. The methods were tested for two gauging stations in Slovenia that exhibit significantly increasing trends in annual maximum (AM) discharge series. The comparison of the models suggests that the stationary model tends to underestimate flood quantiles relative to the non-stationary models in recent years. The model with annual precipitation as a covariate exhibits the best goodness-of-fit performance. For a 10% increase in annual precipitation, the 10-year flood increases by 8%. Use of the model for design purposes requires scenarios of future annual precipitation. It is argued that these may be obtained more reliably than scenarios of extreme event precipitation which makes the proposed model more practically useful than alternative models.