Comparing prediction accuracy of OK and RK for the soils of Surat talukas

Abstract

Prediction of soil properties plays a significant role in forecasting, assessment of risks as well as decision making for Government, agriculturist and other geoscience stakeholders. The acquisition process for these type of data is difficult, time consuming, and expensive. Geographical Information Systems uses several spatial interpolations like Splines, IDW, and Kriging etc. to predict or interpolate unknown environment variables. Kriging belongs to the category of geostatistical interpolation techniques. The major emphasis of this paper is on ordinary kriging which is a method based on weights and regression kriging which is a hybrid method of geo-statistics. This paper compares ordinary kriging that with regression kriging by testing soils of three talukas of Surat district namely Bardoli, Mandvi and Umarpada. Data related to soil major nutrients and micro nutrients have used for comparison. The prediction accuracy of regression kriging is almost 100% as compared to ordinary kriging in three cases taken as sample. Thus the paper signifies the use of kriging techniques for predicting soil properties for these three talukas of southern Gujarat.