The Uniformization of Annual Maximum 1・2・3 Day Precipitations and its Nonstationary Frequency Analysis

This study put in operation the nonstationary frequency analysis by using the uniformized annual maximum 1、2、3 day precipitations。The summary of research results is as follows.1。The approximation of parameter time series(1)The annual maximum1·2·3day precipitations of front，typhoon，cyclone and non-separation for 29years from1964to1992 were used in Nagano，Matsumoto，Iida，Suwa and Karuizawa stations as real data。These precipitation time series had the nonstationarity.(2)When time expresses t，Gumbel distribution，which is assumed as the population distribution，has the nonstationary scale parameter a(T)and location parameter b(T)。The time series of these parameters obtained by 5year moving partial data(It Is Called Parameter Time Series)were approximated by the periodic component and the trend component.(3)The nonstationary parameter time series of Gumbel distribution were approximated by the logistic curve similarly.2。The estimation of nonstationary hydrological variable(1)The hydrological variables of exceedance by using the parameter time series in1.(2)were estimated，and the time series of hydrological variable were drawn.It had the nonstationality.(2)The hydrological variables of exceedance by using the parameter time series in1.(3)were estimated，and the time series of hydrological variable were drawn similarly.It had the nonstationality too.3。The examination of the stability of parameter time seriesThe length of moving partial data was5years in this study.This stability was examined by simulation.The result said that5year moving partial data did not give the sufficient estimated value of parameter time series。

This study put in operation the nonstationary frequency analysis by using the uniformized annual maximum 1・2・3 day precipitations. The summary of research results is as follows.1. The approximation of parameter time series(1) The annual maximum 1・2・3 day precipitations of front, typhoon, cyclone and non-separation for 29 years from 1964 to 1992 were used in Nagano, Matsumoto, Iida, Suwa and Karuizawa stations as real data. These precipitation time series had the nonstationarity.(2) When time expresses t, Gumbel distribution, which is assumed as the population distribution, has the nonstationary scale parameter a (t) and location parameter b (t). The time series of these parameters obtained by 5 year moving partial data (it is called parameter time series) were approximated by the periodic component and the trend component.(3) The nonstationary parameter time series of Gumbel distribution were approximated by the logistic curve similarly.2. The estimation of nonstationary hydrological variable(1) The hydrological variables of exceedance by using the parameter time series in 1.(2) were estimated, and the time series of hydrological variable were drawn. It had the nonstationality.(2) The hydrological variables of exceedance by using the parameter time series in 1.(3) were estimated, and the time series of hydrological variable were drawn similarly. It had the nonstationality too.3. The examination of the stability of parameter time seriesThe length of moving partial data was 5 years in this study. This stability was examined by simulation. The result said that 5 year moving partial data did not give the sufficient estimated value of parameter time series.