Comparison of spatial interpolators for variability analysis of soil chemical properties in Cuamba (Mozambique)

The knowledge of spatial distribution of soil attributes, particularly chemical ones, which is very relevant for agricultural planning. Several studies have focused on spatial interpolation of soil properties, but only a few of them have been undertaken in sub-Saharan Africa. This study aims to analyse the spatial variability of hydrogen potential (pH) and electric conductivity (EC) within an agricultural region in Cuamba district of Mozambique. Efficiency of a deterministic and a stochastic interpolator were compared, namely Inverse Distance Weighting (IDW) and Ordinary kriging, respectively. Soil samples were collected at random locations scattered through the study region, and
were later analyzed in water and soil laboratory. These point data were then used to produce interpolated surfaces of soil chemical properties. Efficiency of spatial interpolation methods was assessed based on prediction errors’ statistics derived from cross-validation. Results show that ordinary kriging was less biased and more accurate than IDW at samples’ locations. Hence, maps produced using the former method are a valuable contribution for the spatial characterization of soil quality, according to its chemical properties. Considering the spatial patterns of pH, southeast area is characterized by clayey soils, which has a high fertility potential for food crops.