SPATIAL VARIABILITY OF SOIL PROPERTIES IN SITIUNG, WEST SUMATRA, INDONESIA (GEOSTATISTICS, SEMI-VARIOGRAM, KRIGING)

Recent developments in statistical theory enable quantitative analysis of spatial dependence and its use for optimal interpolation of soil properties at unsampled locations. Such analysis is achieved by application of geostatistics. This study used geostatistical procedures to augment the traditional soil classification approach for interpolating soil chemical and textual properties in Sitiung, West Sumatra, Indonesia.;Anisotropic spatial dependence of textural components, pH and HCl-extractable P was caused by directional differences in parent material deposition and soil weathering. Isotropic variation and short ranges of spatial dependence for C, N, soil acidity and exchange characteristics suggested the influence of different soil processes affecting the spatial distribution of different properties.;Soil properties were interpolated for unsampled locations by punctual, block and co-kriging values from measured data locations. Estimation variances of each kriged value indicated the reliability of interpolation. Punctual kriging based on anisotropic semi-variogram models gave more precise estimation than when isotropic models were used for interpolation of textual components, pH and HCl-P. Block kriging of lime requirement and total N over cells of 4-km('2) gave smaller estimation variances and smoother isarithm maps compared to punctual kriging. Co-kriging reduced estimation variances (relative to auto kriging) of available P in areas where it was undersampled relative to the covariable, reserve P.;Geostatistical analysis of soil properties in a 0.1 ha plot suggested that local effects of land clearing were superimposed on the long range depositional and weathering processes in the region. This suggests that soil variability is caused by processes operating and interacting over different spatial scales. Growth and yield of upland rice was greater on burn patches within the plot than on areas of exposed subsoil and the surrounding soil. High concentration of exchangeable Al was the main chemical constraint to crop growth in the latter two areas.;Multiple regression and hierarchical analysis of variance indicated that the order and modifier levels of Soil Taxonomy and the Fertility Capability Classification, respectively, accounted for most of the variance of organic C, total N, P, soil acidity and exchange characteristics in the Sitiung region.;Results demonstrate that geostatistics can improve understanding of soil-forming processses and can augment classical interpolation methods by estimating soil properties with known precision.