Evaluating accuracy some of geostatistical methods to predict spatial variability of final infiltration rate

Soil properties continuously experience spatial variations in time and space. An awareness of these changes is necessary for increase profitability and sustainable agriculture management. In this study geostatistical techniques were used to describe the spatial structure of final infiltration rate and select an effective interpolation method to predict infiltration rate. From the study area 75 soil samples were collected by a systematic sampling strategy on gird spacing nearly 10 meter at 0 to 25 cm depth. Infiltration rate was measured in the field by using double-ring infiltrometeres until steady state (IRs). Three interpolation methods were used to estimate (IRs), which included inverse distance weight (IDW) with power 1 to 3, kriging and cokriging with Auxiliary variable, that including silt content and organic matter. The statistics mean absolute error (MAE) and mean bias error (MBE) with cross validation techniques were used to compare the prediction quality. Results showed that, the coefficient of variation of soil properties ranged from 7.26 % (silt content) to 88.49% (IRs) at this field. Result semivariogram showed that infiltration rate had range 24 m. According to the statistical parameter, both kriging and cokriging provided reasonable estimation as compared IDW. However using silt content as auxiliary data for estimating of IRs in cokriging method could reduce prediction error by 14% as compared kriging method.