Comparison of Homogenous Region Delineation Approaches for Regional Flood Frequency Analysis at Ungauged Sites

Abstract

The occurrence of floods and their impacts on hydrologic systems and society are critical considerations in the design and management of a large number of water resources projects. As streamflow records are often limited or unavailable at many watersheds, it is necessary to develop better methods for regional estimation of floods at these partially-gauged or ungauged sites. In this paper, a comparative study of homogeneous region delineation approaches in the context of regional flood quantile estimation is presented. Three approaches were considered in this study using the available annual maximum series (AMS) of flood peaks for 57 watersheds in Québec, Canada. The first approach was the scaling method, which is a relatively new approach for homogeneous region delineation. This approach is based on the scaling behavior of the flood series with the basin area. The other two approaches were the region of influence (ROI) and canonical correlation analysis (CCA) methods. Regional flood estimation was carried out using the index flood method. The jackknife technique was used to represent the ungauged site condition. Analyses of hydrologic data from Québec have indicated that the AMS exhibited a simple scaling behavior and that the drainage area alone may describe most of the variability in the statistical moments of the AMS data. Results of this numerical application have found that the flood quantile estimates obtained from the scaling approach are more accurate and more robust than those given by the ROI or CCA methods. In addition, it was found that the grouping of similar basins on the basis of the scaling approach formed well-defined geographical regions with distinct climatic characteristics.