Abstract
Abstract. Hydraulic models for flood propagation description are an essential tool in many fields and are used, for example, for flood hazard and risk assessments, evaluation of flood control measures, etc. Nowadays there are many models of different complexity regarding the mathematical foundation and spatial dimensions available, and most of them are comparatively easy to operate due to sophisticated tools for model setup and control. However, the calibration of these models is still underdeveloped in contrast to other models like e.g. hydrological models or models used in ecosystem analysis. This has two primary reasons: first, lack of relevant data against which the models can be calibrated, because flood events are very rarely monitored due to the disturbances inflicted by them and the lack of appropriate measuring equipment in place. Second, 2-D models are computationally very demanding and therefore the use of available sophisticated automatic calibration procedures is restricted in many cases. This study takes a well documented flood event in August 2002 at the Mulde River in Germany as an example and investigates the most appropriate calibration strategy for a simplified 2-D hyperbolic finite element model. The model independent optimiser PEST, that enables automatic calibrations without changing model code, is used and the model is calibrated against over 380 surveyed maximum water levels. The application of the parallel version of the optimiser showed that (a) it is possible to use automatic calibration in combination of 2-D hydraulic model, and (b) equifinality of model parameterisation can also be caused by a too large number of degrees of freedom in the calibration data in contrast to a too simple model setup. In order to improve model calibration and reduce equifinality, a method was developed to identify calibration data, resp. model setup with likely errors that obstruct model calibration.
Highlights
Floods are serious events and may have severe socioeconomic impacts on vulnerable areas
In this case more meaningful indexes for the comparison of the different calibrations are, besides the objective function used in PEST: the mean absolute error (MAE), the root mean square error (RMSE) and the average error (BIAS) of the simulation results from the measured maximum inundation depths calculated as follows flood depths were surveyed by laser reflectometry of water marks on buildings above ground at 390 points, yielding detailed point information of inundation depths in the town
Except for the roughness coefficient related to the channel, each calibration gave quite low Strickler roughness values, i.e. high hydraulic resistance, for the floodplain compared to literature values
Summary
Floods are serious events and may have severe socioeconomic impacts on vulnerable areas. Gradient-based methods on the other hand are computational very efficient but the solution can be dependent on the initial parameter values and they might get trapped in a local minimum, if the response space of the objective function is highly complex These methods may be the only possibility to automatically calibrate CPU time demanding models, like the one presented here. In this study area the flood event occurred in August 2002 was well documented: flood depths were recorded from a large number of water marks, the maximum inundation extent was surveyed from satellite imaging and water marks, and the flood hydrograph was recorded at the upstream flowgauge This data set enabled the automatic calibration, it will show that a large amount of data or information do not assure an improvement in the identification of the parameters. In this paper a procedure to remove potentially erroneous data is presented
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
