平均値と分散のジャンプ検出による確率水文量の更新

-北海道内22気象官署の年最大日雨量と年最大無降雨連続日数への適用-

抄録

There is a possibility that the population of extreme precipitation has been changed by a climate change. Therefore it is necessary to estimate the probable hydrological amount which can correspond to the population change using an unsteady frequency analysis.
In this paper, each precipitation data set of 22 meteorological observatories in Hokkaido is divided into 2 partial-duration time series of sample size, n₁ and n₂ (n₁≥7, n₂≥30, N=n₁+n₂), to detect a mean and a variance jump. Then, T-year probable hydrological amount using n₂ data which is significant for the jump is estimated to compare with the predetermined probable hydrological amount for complete-duration series of sample size, N. Consequently, we clarify that T-year probable hydrological amount of the annual maximum daily rainfall and the annual maximum number of days with continuous non-rainfall need to be renewed by 0.1-49% increase for 12 observatories and 2-14% increase for 5 observatories respectively.