抄録
Examination and estimation of design flood discharges in upper river basins are critical for better river and erosion control planning. Annual maximum flood series were analyzed by 13 fitting methods in terms of SLSC goodness-of-fit criterion at 30 small-scale [mostly less than 100 km2] dam reservoirs in Hokuriku region where inflows are influenced by snow-melting but deemed unregulated. The inflow discharge data belong to the same hydrological regime and show mutual independency and proper randomness as a set of statistical sample. The analysis of fitting methods show that log Pearson Type III distribution is the best common probabilistic distribution for the sample annual maximum flood series in the region, which allowed us to apply regional flood frequency analysis method described in the U.S. Water Resource Council guideline Bulletin 17 B. 30 dam reservoirs are located within about one-hundred-mile radius and their flood series are pooled for the regional flood frequency analysis. Confidence intervals were estimated to become significantly smaller as regional information was incorporated than when estimated separately at each site. Concurrent omparative analysis by an index-flood method and a scaling method indicated that higher statistics such as skew coefficient at each observing station are indispensable in addition to mean, standard deviation, and coefficient of variation for further improvement of the estimation. Bulleting 17 B method, which adjusts site skew by regional information, seems promising to apply to small-scale non-regulated river basins in Hokuriku region. Further research is necessary to clarify how river basin characteristics are related to site skew in maximum annual flood series.
