Homogeneous pooling group delineation for flood frequency analysis using a fuzzy expert system with genetic enhancement

Pooled flood frequency analysis has been widely used to improve flood quantile estimation at sites that have short streamflow gauging records. When conducting pooled frequency analysis, it is necessary to delineate homogeneous pooling groups. A basic requirement for a pooling group is that the catchments in a pooling group should have sufficient similarity in hydrological response to provide a basis for information transfer. In this paper, methods using fuzzy expert systems (FES) are developed to derive an objective similarity measure between catchments. The steps for building the FES are explained in detail. The performance of the FES is improved by tuning of the membership functions of the fuzzy sets using a genetic algorithm. The proposed methods are applied to flood data from Great Britain and compared with results obtained from two other methods. The results indicate that the proposed FES methods have the capability of utilizing inputs from different categories of catchment characteristics and producing similarity measures that have high reliability.