Journal of Watershed Management Research

Regional analysis is the stability method to improve estimates of flood frequency, which has become one of the dynamic sectors in hydrology and the new theories are testing, constantly. Application of geostatistical method is an innovation in this field for regional flood analysis.This technique is based on the interpolation of hydrological variables in the physiographical space instead of using common geographical space. However, the hydrological, climatological and physiographical information of 38 river basin information of watersheds inside of Mazandaran province were obtained and Canonical Kriging method was used for regional flood analysis whit return periods of 10, 20, 50 and 100 years. The Canonical Correlation Analysis was used to design physiographic space by the geomorphoclimatical variables affecting the flood. The Gaussian model had the best fit to semi-variogram model in all return periods and the geostatistical method of Ordinary Kriging was used for regionalization. Results evaluated using Jack Knife mutual procedure and the five statistical indexes. The performance of NASH was obtained over than 0.9 in all of return periods, which implies the accurate and acceptable predictions of flood in ungauged basins. The other indicators led to satisfying results, too. According to the results of relative statistical indicators accuracy regional estimates improves as long as return periods increase. These results showed that application of Canonical Kriging method is effective and practical approach for regional flood analysis.