A spatial-homogeneity-based interpolation algorithm for soil properties

Estimation of soil properties in an un-sampled location is generally calculated with observation values of a number of neighbors around the un-sampled location to reduce the computational complexity. The more similarity there is between the neighbors, the better prediction at the un-sampled location. In this paper, we proposed a spatial homogeneity-based interpolation algorithm, which finds more similar neighbors close to an unmeasured location to improve estimation performance. The algorithm imposes a constraint on interpolation process, which the estimation of soil properties at unknown locations is processed in the context of spatial homogeneity. The experimental results on real data of 196 soil Cu illustrates that the performance of the algorithm is more reliable than OK.