Investigation of the maximisation potential of large floods by spatio-temporal simulation of rainfall

Exceptional floods are caused by extreme meteorological conditions matching critical space-time scales of flood generation processes in a catchment. Although these events are very rare, they are possible and it would be helpful to know what might happen in these cases for risk assessment. Fortunate, for most of the flood events some factors are not “optimal” to cause an exceptional flood. This study investigates which reasonable changes on the flood producing storms may lead to exceptional floods. For that, observed storms are stochastically modified and analysed regarding its maximisation potential for floods. First the spatial patterns are modified, then the intensities are modified and finally both conditions are modified together. The pattern modification is based on space-time simulation conditioned on observations from a conventional rainfall station network. The rainfall intensities are modified with moisture maximisation but keeping the observed pattern constant according to radar rainfall observations. Many space-time rainfall realisations are produced and used as input for a rainfall-runoff model. Simulated annealing is used as conditional rainfall simulation approach and the HBV model to simulate the floods. This case study uses data from the Mulde river catchment in Germany and applies the methodology to a set of selected large flood events. It is expected, that the results reveal how extreme the floods could have become and whether increased intensities or pattern modification contributes more to the maximisation potential.