Abstract
This paper describes regionalization of probable rainfall that plays important role in water resources planning, especially in planning of flood prevention facilities. The authors compare some procedures for obtaining estimates of probable rainfall at arbitrary locations that have no observation. The procedures are: (1) simple mm regional regression analysis, (2) stationary kriging, (3) nonstationary (universal) kriging, and (4) regression-kriging combined analysis. Applying these procedures to the Yasu River basin (387km2) in Japan, they have obtained the estimates and the estimation accuracy for probable rainfall, compared them, and discussed incorporation of orographic effects.
