Regional flood frequency analysis by combining self-organizing feature map and fuzzy clustering

Regionalization is the procedure to find natural groups of watersheds with homogeneous hydrologic response, and finds applications in hydrologic design, planning and management of water resources systems. In regionalization studies, clustering techniques are useful to partition catchments in a region into natural groups. The linear Kohonen’s self-organizing feature map (SOFM) has been applied as a clustering technique for regionalization in several recent studies. However, SOFM is not a clustering method because it is seldom possible to interpret clusters from the output of an SOFM, irrespective of its size and dimensionality. In this study, we demonstrate that SOFMs may, however, serve as a useful precursor to clustering algorithms. We present a two-level SOFM-based clustering approach for regionalization of watersheds. In the first level, the SOFM is used to form a two-dimensional feature map. In the second level, the output nodes of SOFM are clustered by using Fuzzy c-means algorithm to form regions for flood frequency analysis (FFA). The optimal number of regions is based on fuzzy cluster validation measures. Effectiveness of the proposed approach in forming homogeneous regions for FFA is illustrated through application to data from watersheds in Indiana, USA. We find that previous indices used to decide the number of clusters are not efficient, and therefore suggest a new index that is more appropriate for this purpose. Using leave-one-out cross-validation, the performance of the proposed approach to regional FFA is compared with those from methods based on regression analysis and canonical correlation analysis. Results show that the proposed approach performs better in estimating flood quantiles at ungauged sites.