Application of the stochastic self-training procedure for the modelling of extreme floods

Modelling techniques valid for ordinary hydrological conditions are often inadequate for catastrophic runoff conditions, as the majority of the parameters included in the modelling schemes are not necessarily applicable to flood conditions. To overcome this problem, using a stochastic self-training procedure (SSTP) is proposed. The procedure is based on the real water balance equation and the statistical model of the water balance. An application of the procedure with data from the Filyos River (Turkey) is presented in this study.