Article ID: | iaor19962185 |
Country: | South Africa |
Volume: | 29 |
Start Page Number: | 65 |
End Page Number: | 78 |
Publication Date: | Nov 1995 |
Journal: | South African Statistical Journal |
Authors: | Waal D.J. de, Groenewald P.C.N. |
Keywords: | statistics: inference |
A Bayesian estimation procedure is suggested as an alternative to the well-known kriging method in estimating ground-water levels. The Bayesian method uses what Omre called ‘qualified guesses’ as prior information. The knowledge of the topography is used to determine qualified guesses in this study and a comparison of these Bayes estimates with various kriging estimates shows much smaller errors. The procedure proposed here differs from that given by Omre in that no weights have to be calculated through minimization of the variance of the estimator.