抄録
Probable maximum precipitation (PMP) is one of the keys for the designing of probable maximum flood (PMF) and flood disaster management. A basin-scale spatial distribution analysis of the extreme rainfall and PMP in the Yodo river basin are presented. The maximum 24-hour rainfall data from 1881 to 2011 were used. Spatial PMP and its isohyetal lines produced using Hershfield and Spline interpolation method agrees with the historical flood records. Highest PMP is observed around the Hikone and Shimogahara stations (above 900 mm), while the least at Yanagase and Torahime (below 500 mm). Estimation of the return periods (RPs) using non-linear regression and the generalized extreme value (GEV) distribution for the historical maxima suggests that the GEV overestimates RPs and non-linear regression (NLR) with PMP as upper bound stabilizes the estimates of RP by GEV. Spatial distributions of the 4000 years return periods extreme rainfalls show some similarities with the PMP's distribution.