Application of geostatistics for grid and random sampling schemes for a grassland in Nigde, Turkey.

Abstract

Soil quality prediction maps are important tools for environmental scientists and policymakers. An 18ha grassland was selected to create soil quality prediction maps. A total of 30 sampling points were selected, and samples were collected from top soil (0-20cm depth). Twelve of the sampling points were selected randomly and 18 of the sampling points were selected based on a square shaped grid plan. The soil samples were then analyzed for pH, electrical conductivity (EC), organic matter (OM), water content, dissolved total carbon (DTC), dissolved organic carbon (DOC), dissolved inorganic carbon (DIC), and dissolved total nitrogen (DTN). Ordinary kriging (OK) and ordinary cokriging (OCK) spatial interpolation methods were used for the prediction of spatial distribution. The prediction errors showed that the parameters in the grid sampling scheme showed better prediction in the OK technique. The highest reduction in prediction errors was obtained in the DOC in grid sampling scheme after using OCK.