This master’s thesis studies the optimization of the rainfall input that is used for the modelling of the Demer basin and river. This thesis will evaluate whether radar images provide additional value to the rainfall input used in these hydrological models. The radar images that will be used are from the weather radar of Wideumont and Zaventem. An analysis is conducted for the months of November 2010 and January and August 2011. The radar images will be adjusted by the pluviographs by means of the simple kriging method. To evaluate this method, a comparison is made between the registered rainfall at different pluviographs, the corrected radar images and the original radar images. The conclusion of the first part of this thesis is that the precipitation based on the original radar images produces inferior results. The corrected radar images produce the best results for the amount of the precipitation and the time of the precipitation.
In a second part the output of the hydrological model is compared with the measured flow rate. After making an analysis of the results, it can be concluded that the input, based on the corrected radar images, leads to improved results compared to results based on the pluviograph data. For the months of November and January, only a small improvement is made in the modelling of the flow rate. On the other hand, the results of the modelled flow rate for August clearly show that the corrected radar images provide better estimations of these flow rate measurements.
The third part of this thesis consists of the hydrodynamic modelling of the Demer. The input for this model are the run off flow rates that were simulated with the hydrological model. The output from this model consists of the water level and the flow rate of the Demer. These results are compared with measurements of the water level and flow rate and again there can be concluded that there is a small improvement for the results for the months of November and January, but there is a significant improvement for the month August.
Finally the last section looks at the possibility of making precipitation forecasts by extrapolating radar images over time. There are two different methods considered in this thesis. Method A uses a distance based relation between the cores of the precipitation zone in the radar images at time (t-1) and time (t), to extrapolate the radar image. The second method is based on the two dimensional cross correlation between the radar image at time (t-1) and time (t). After making a comparison between the forecasted radar images and the registered and corrected radar images, which are used for reference, method A combined with the radar of Wideumont results in the best forecast for the basin precipitation. Method B combined with the radar of Wideumont makes the best estimation of the direction of the precipitation zones. No conclusion can be made regarding the evolution of the error when expanding the forecasted time window.
