Several methods can be used to study the spatial variability of soil organic carbon at large scales. In this study, some interpolators were assessed for smoothing soil organic carbon (SOC) contents in 100 sites study in the Azna watershed, Iran. The employed interpolation techniques were consisted of inverse moving average, moving average, minimum curvature, nearest neighborhood, ordinary Kriging and universal Kriging. The precision of interpolation methods was measured by cross validation, using the mean absolute error (MAE) between the interpolated and observed values. The experimental variogram of SOC was then fitted to an exponential model. The results indicated that the MAE values were ranged from 0.212 to 0.368 percent depending on the method and the number of neighbors used for interpolation. The obtained results also indicated that the minimum curvature and moving average methods were the most precise methods to predict the SOC contents. The “inverse moving average” appeared to be the less precise method, for which the MAE values were 0.212, 0.220 and 0.368 percent, respectively. Although the MAE values for the two used techniques were similar, but the interpolated surfaces were different. It was therefore concluded that due to its precision for interpolation and the smoothness of the interpolated surface, the “minimum curvature” was the most appropriate method for smoothing SOC.
