WAVELETS FOR SYNTHESIS OF MEAN DAILY FLOWS FROM MONTHLY VALUES

摘要

In this paper a methodology for generation of missing daily flow records is developed through finding correlation between high frequency wavelet coefficients of series of daily flows recorded at gage stations with similar hydrological behavior. Using Haar wavelet, level 5 decomposition approximations can be taken as the mean monthly values; if normality assessment for wavelet coefficients is assumed the detail coefficients are completely characterized by its mean and variance. Knowing daily data at one gage station, wavelet decomposition and normalization of high frequency coefficients for different levels are attained. The mean and standard deviation of wavelet coefficients, for each level, at site with missing data are obtained through correlation with same statistics from other gage stations; the estimated values are used for reversing normalization so details coefficients are synthesized and daily flow values can be obtained for the missing period. Procedure is verified for four gage stations at Ebro basin through 3 year length data series from 1969 to 1999 period.