Mapping the conditional probability of soil variables

Abstract

In a study area of 410 ha, located in a region with acid sulphate soils in Indonesia, 790 soil observations were made at a regular grid of 75 m × 75 m. Additionally, 74 test observations were made, randomly distributed over the area. The variable “depth to the pyritic layer” was selected for this study. Two different methods were presented for the production of conditional probability maps of this variable. The first method is based on the use of kriging. The second is based on the use of the inverse distance spatial prediction technique and test points. Both techniques were used to predict values of the variable for a grid of 25 m × 25 m. Besides predicted values for every grid point, an estimation of the prediction error, expressed as the standard deviation, was also made. Kriging produces this standard deviation automatically. With the inverse distance technique the standard deviation was estimated using test borings. The standard deviations obtained were used to calculate confidence intervals for each grid point. Results are presented in maps showing the probability that the depth to the pyritic layer exceeds 50 cm. Only small differences were found between the probability maps produced using the two procedures.