SPATIAL DISTRIBUTION OF INFILTRATION PARAMETERS AT REGIONAL SCALE

Spatial variability of infiltration parameters were investigated in about 4000 ha of Kavar plain, east of Shiraz, Iran. For this purpose, infiltration was measured on a systematic squared grid pattern with 500 by 500 m in 161 sites using double ring infiltrometer. Soil bulk density (BD), initial soil water content, soil texture, organic carbon content (OC), saturation percentage (SP), pH, and EC were also measured. The interpolation of infiltration parameters in non-sampled areas were predicted using kriging, inverse distance weighted, and co-kriging methods. The experimental semi- variograms were fitted to linear, exponential, gaussian, and spherical models. Cross validation method using statistical parameters of MBE, MAE , MSE, and RMSE was used to choose the most accurate interpolation method. The spatial distribution map of parameters was obtained using the best interpolation method. Estimating the coefficient of a in Kostiakov model using co-kriging method (spherical model) and co-factor of BD with R 2 =0.99, nugget effect of -0.002, and range of 3380 m was superior to kriging method. The power b of Kosiakov was estimated by exponential mode and kriging method with R 2 = 0.97, nugget effect of 0.005 and range of 14410 m. To estimate sorptivity parameter (S) of Philip model, the co-kriging method with spherical model and co-factor of BD with R 2 =0.99 and range of 2973 m was preferred to kriging method. Likewise, estimation of transmissivity parameter of Philip model using co-kriging method (Gaussian model) with R 2 =0.996, nugget effect of 0.0003, and range of 3862 m showed high accuracy. Using soil moisture content, BD, texture, SP, and organic carbon as co-factors of co-kriging method, resulted in relatively better estimation of infiltration model parameters.