河川의 日別水溫差에 對한 推計學的特性

Stochastic Properties of Daily Temperature in Rivers

The stochastic characteristics of the daily range of water temperature variation was analyzed by employing the techniques of autocorrelation coefficient, autoreggresive model and crosscorrelation model. These time series inc1uded dai1y observations on maximum and minimum values of water temperature and air temperature. The measurement was made.by automatic recording instrument at Gu-yee and Dook-<io in Han River, and at Waegwan and Gu-mi in Nackdong River in 198 1. As a result of this study, it was found that (1) The correlogram of daily water temperature ranges AAT j and daily air temperature l:!,ATj at Gu-mi and Gu-yee showed the exponential curves. (2) The most high frequency valuesof l:!,ATj and l:!,WTjwere l1 Y C and O.5 Y C respectively at every measuring site. (3) The correlation coefficients between the daily mean air temperature ATj and the dai1y mean water temperature were fairly high as 0.966 at Dook-do and 0.949 at Gu-yee, but the correlation coefficients between l:!,ATj and l:!,WTj were very low as 0.1074 at Gu-yee and 0.0324 at Dook-Do. (4) It was analyzed that the most adequate autoregressive model of the four types were 父j + 1 v ~ _ 2 \Y.z_ ... ____ _1 __ -...:1_ C. _ ,7 (I - PI)X + p, Xj + P (I - p, ) Ej type at Gu-yee and Dook。， and Xj + 1 =(l -p,)X+ p 1 Xj +P (1 一p, 2)V2 Ej type at Waegwan and Gu-mi. (5) The most adequate simple regressive model of the four types for predicting l:!,WT j was Y =(1 - r)Y + r Xj + (1 - r2 )V2 Ej at Gu-yee and Dookφ in the view of matching the vz probability denslty function of 0nal time series, and Y = (1 - r)Y + r ~ Ej type or Y =(1 - r)Y + ‘ + 0, (1 - r2 )1/2 type at each site in the view of closing to the mean value of UX the original time series.