Application of geostatistical techniques for in situ site characterization has received much attention in the last couple of decades. Kriging is one of the popular and accurate geostatistical techniques for the interpolation of spatially varied random variables. This study considers the development of a generalized ordinary kriging algorithm for use with site characterization. Ordinary kriging with a constant mean of stationary variable was applied to generate contour map and the error variance map to infer on the spatial variation of the parameter under consideration. In this study, the clay content parameter from a refinery project area in Orissa is interpolated using ordinary kriging technique. A generalized MATLAB code is developed to select the best fit semi-variogram for the sample data, to apply ordinary kriging technique, and to generate the surface profile. The spatial distribution of clay content values across the region is studied using prediction surface, and accuracy is checked using error variance profiles. Results of the analysis are also compared with simulation using ArcGIS based geostatistical analyst® and cross-validated using statistical parameters. Our results conclude that the proposed algorithm can be extended to predict other in situ soil properties in the field of geotechnical engineering. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.