Mulde Catchment Research Articles

How do changes along the risk chain affect flood risk?

Abstract. Flood risk is impacted by a range of physical and socio-economic processes. Hence, the quantification of flood risk ideally considers the complete flood risk chain, from atmospheric processes through catchment and river system processes to damage mechanisms in the affected areas. Although it is generally accepted that a multitude of changes along the risk chain can occur and impact flood risk, there is a lack of knowledge of how and to what extent changes in influencing factors propagate through the chain and finally affect flood risk. To fill this gap, we present a comprehensive sensitivity analysis which considers changes in all risk components, i.e. changes in climate, catchment, river system, land use, assets, and vulnerability. The application of this framework to the mesoscale Mulde catchment in Germany shows that flood risk can vary dramatically as a consequence of plausible change scenarios. It further reveals that components that have not received much attention, such as changes in dike systems or in vulnerability, may outweigh changes in often investigated components, such as climate. Although the specific results are conditional on the case study area and the selected assumptions, they emphasize the need for a broader consideration of potential drivers of change in a comprehensive way. Hence, our approach contributes to a better understanding of how the different risk components influence the overall flood risk.

Spatially coherent flood risk assessment based on long-term continuous simulation with a coupled model chain

Summary A novel approach for assessing flood risk in river catchments in a spatially consistent way is presented. The approach is based on a set of coupled models representing the complete flood risk chain, including a multisite, multivariate weather generator, a hydrological model, a coupled 1D–2D hydrodynamic model and a flood loss model. The approach is exemplarily developed for the meso-scale Mulde catchment in Germany. 10,000 years of meteorological fields at daily resolution are generated and used as input to the subsequent models, yielding 10,000 years of spatially consistent river discharge series, inundation patterns and damage values. This allows estimating flood risk directly from the simulated damage. The benefits of the presented approach are: (1) in contrast to traditional flood risk assessments, where homogenous return periods are assumed for the entire catchment, the approach delivers spatially heterogeneous patterns of precipitation, discharge, inundation and damage patterns which respect the spatial correlations of the different processes and their spatial interactions. (2) Catchment and floodplain processes are represented in a holistic way, since the complete chain of flood processes is represented by the coupled models. For instance, the effects of spatially varying antecedent catchment conditions on flood hydrographs are implicitly taken into account. (3) Flood risk is directly derived from damage yielding a more realistic representation of flood risk. Traditionally, the probability of discharge is used as proxy for the probability of damage. However, non-linearities and threshold behaviour along the flood risk chain contribute to substantial variability between damage probabilities and corresponding discharge probabilities.

PAI-OFF: A new proposal for online flood forecasting in flash flood prone catchments

Summary The Process Modelling and Artificial Intelligence for Online Flood Forecasting (PAI-OFF) methodology combines the reliability of physically based, hydrologic/hydraulic modelling with the operational advantages of artificial intelligence. These operational advantages are extremely low computation times and straightforward operation. The basic principle of the methodology is to portray process models by means of ANN. We propose to train ANN flood forecasting models with synthetic data that reflects the possible range of storm events. To this end, establishing PAI-OFF requires first setting up a physically based hydrologic model of the considered catchment and – optionally, if backwater effects have a significant impact on the flow regime – a hydrodynamic flood routing model of the river reach in question. Both models are subsequently used for simulating all meaningful and flood relevant storm scenarios which are obtained from a catchment specific meteorological data analysis. This provides a database of corresponding input/output vectors which is then completed by generally available hydrological and meteorological data for characterizing the catchment state prior to each storm event. This database subsequently serves for training both a polynomial neural network (PoNN) – portraying the rainfall–runoff process – and a multilayer neural network (MLFN), which mirrors the hydrodynamic flood wave propagation in the river. These two ANN models replace the hydrological and hydrodynamic model in the operational mode. After presenting the theory, we apply PAI-OFF – essentially consisting of the coupled “hydrologic” PoNN and “hydrodynamic” MLFN – to the Freiberger Mulde catchment in the Erzgebirge (Ore-mountains) in East Germany (3000 km 2 ). Both the demonstrated computational efficiency and the prediction reliability underline the potential of the new PAI-OFF methodology for online flood forecasting.

Flash flood forecasting combining meteorological ensemble forecasts and uncertainty of initial hydrological conditions

Flood forecasting for fast responding catchments encounters problems, especially in terms of short warning periods and a very limited reliability. Within a new stochastic framework based on rigorous rainfall-runoff modelling and Monte Carlo simulations, we consider uncertainties of two sources: (i) uncertainty from the estimation of initial hydrological conditions, and (ii) the uncertainty of the meteorological rainfall forecast. We avoided the high computational demand of extensive Monte Carlo simulations by using a symbiosis between physically-based hydrological modelling and computationally highly efficient artificial intelligence techniques. The new PAI-OFF methodology (Process Modelling and Artificial Intelligence for Online Flood Forecasting; see Schmitz et al, 2005, and Cullmann, 2006) employs a physically-based hydrological/hydraulic model of the considered catchment for generating, in a first step, the complete range of realistic possible flood scenarios on the basis of a catchment specific meteorological analysis. The resulting database of corresponding input/output vectors – supplemented by generally available hydrological and meteorological data for characterising the catchment situation prior to a storm event – serves, in a second step, for setting up a set of task-specific artificial neural networks (ANN), which finally portray both the rainfall-runoff process and the hydrodynamic flood wave propagation in the river. We subsequently use this tool for investigating the global uncertainty of flash flood forecasting in a small-to medium-sized catchment on the basis of a comprehensive Monte Carlo analysis. Along these lines, the computationally highly efficient PAI-OFF technique allowed performing an extensive number of runs for obtaining ensembles of predicted stream flow that can be used to evaluate probabilities of exceedance of critical river stages/flows via an integration of both the hydrological uncertainty and the meteorological uncertainty. This approach was then implemented and applied to the Freiberger Mulde catchment in the Ore Mountains in Eastern Germany (with an area of about 3000 km2). The results of the overall ensemble predictions in the form of the ensemble mean values unveiled an astonishing underestimation of the recorded flood peak – most due to the bias of the considered initial hydrological conditions.

Aspects of seasonality and flood generating circulation patterns in a mountainous catchment in south-eastern Germany

Abstract. Analyses of discharge series, precipitation fields and flood producing atmospheric circulation patterns reveal that two governing flood regimes exist in the Mulde catchment in south-eastern Germany: frequent floods during the winter and less frequent but sometimes extreme floods during the summer. Differences in the statistical parameters of the discharge data can be found within the catchment from west to east. The discharges are compared to a number of landscape parameters that influence the discharge in the sub-catchments. Triggering circulation patterns were assigned to all events of the annual maximum discharge series in order to evaluate which circulation patterns are likely to produce large floods. It can be shown that the cyclone Vb-weather regime (TM, TRM) generates the most extreme flood events in the Mulde catchment, whereas westerly winds produce frequently small floods. The Vb-weather pattern is a very slowly moving low pressure field over the Gulf of Genoa, which can bring large amounts of rainfall to the study area. It could also be shown that even with the two flood regimes estimates with the annual maximum series provide a safer flood protection with a larger safety margin than using summer maximum discharge series for extreme summer floods only. In view of climate change it is necessary to integrate knowledge about catchment characteristics, the prevailing flood regime or the trends of weather patterns in the estimation of extreme events.

Identifying and reducing model structure uncertainty based on analysis of parameter interaction

Abstract. Multi-objective optimization algorithms are widely used for the calibration of conceptual hydrological models. Such algorithms yield a set of Pareto-optimal solutions, reflecting the model structure uncertainty. In this study, a multi-objective optimization strategy is suggested, which aims at reducing the model structure uncertainty by considering parameter interaction within Pareto-optimal solutions. The approach has been used to develop a nested setup of a rainfall-runoff model, which is integrated in a probabilistic meso-/macroscale flood forecasting system. The optimization strategy aided in determining the best combination of a lumped (computationally efficient in operational real time forecasting) and a semi-distributed parameterization of the hydrological model. First results are shown for two subbasins of the Mulde catchment in Germany. The different phenomena of parameter interaction were analysed in this case study to reduce the model structure uncertainties.

Risk assessment and mapping of extreme floods in non-dyked communities along the Elbe and Mulde Rivers

Abstract. Assessing and mapping damage risk of floods for large river basins is still in its infancy. Damage risk is understood to be the combination of flood hazard and the vulnerability of communities to a flood of a particular return period. Risk is calculated and mapped for two communities in which dykes are not located for flood protection: Meissen on the Elbe River and Döbeln in the Mulde catchment. Different methodologies for the computation of flood depth and inundation extent of varying flood return periods (hazard) are compared. Exposure and relative damage to the flooding (vulnerability) based on land-use coverages of different scale are also compared and discussed. A property asset coverage completes the data requirements for the construction of the risk maps. Recommendations for continued research on risk assessments of large river basins conclude the study.

Modelling the runoff response in the Mulde catchment (Germany)

Abstract. The paper presents two different levels of regionalization used to represent the spatial distribution of landscape parameters for the hydrological modelling of the Mulde. The aim of this investigation was to find out how the discretisation level affects quality of modelling with the hydrological modelling system J2000. Furthermore we improved our understanding of the applicability and reliability of the distributed model J2000 on the macro-scale. Spatial information was aggregated in two different discretisation levels: subbasins (SB) and hydrological response polygons (HRP). A J2000 simulation was carried out for both discretisation levels based on a 1 year calibration and a 3 year validation period. Simulations performed well for both levels of spatial discretisation. The results seemed to be better in the more complex discretisation approach, where the Nash-Sutcliffe coefficient was higher. We can conclude that our first results show more accurate simulations produced by the HRP discretisation approach, the visual inspection shows a better application of the SB approach to the reproduction of the base flow.

Proportions of subsurface drainages in large areas—methodological study in the Middle Mulde catchment (Germany)

AbstractHigh nitrogen (N) input to rivers requires measures for the reduction of diffuse N pollution. Beside the groundwater, artificial subsurface drainage systems are the main pathways of diffuse N input into rivers. Nevertheless, the N discharge via subsurface drainage systems is one of the main missing links for modeling, especially because of the lack of data bases of subsurface drainage areas. We introduce a method to calculate the normally unknown proportions of drained areas in arable lands improving the existing method by Behrendt et al. (2000). The method is applied for the catchment of Middle Mulde river (area: 2,700 km2) in Saxony/Germany. The data records of the mesoscale soil mapping are allocated to the subsurface‐drainage areas digitalized in representative areas using ARC/INFO GIS. In this way, it is possible to establish a differentiated record of the proportion of subsurface‐drainage area of each regional site type. The results were extrapolated to the entire area by transferring the proportions of subsurface‐drainage areas to areas where no information on drainage areas was available. The approach is well‐suited for future model approaches on a regional scale. By creating and integrating new data sets derived from modern GIS operations, the approach reduces the uncertainty of modeling N and water fluxes.

Climate change scenarios and runoff response in the Mulde catchment (Southern Elbe, Germany)

The impact of a climate change scenario on regional climate conditions and runoff characteristics has been investigated for the Mulde catchment, a meso-scale sub-basin of the Elbe in Germany. First, the semi-distributed, conceptual model HBV-D has been successfully applied to simulate discharge for present climate conditions. Further, the expanded downscaling method (EDS) was calibrated and applied to observed global circulation fields in order to produce local climate input data for HBV-D. Finally, the coupled atmosphere-ocean model ECHAM4/OPYC3, driven by a climate change scenario, provided simulated global circulation patterns for application with EDS. The regionalised scenario conditions then served as input to HBV-D in order to investigate the impact of global climate change on regional hydrology. The results indicate that an obvious increase in temperature is accompanied by a clear tendency to reduced precipitation over the investigated area for the next 100 years. These conditions lead to a decrease in simulated mean discharges of the Mulde. The study is considered to be a contribution for regional impact studies on global climate change. At the same time, it demonstrates existing shortcomings and limitations of current climate impact research.