Standardized vs. customary ordinary cokriging: Some comments on the article “The geostatistical analysis of experiments at the landscape-scale” by T.F.A. Bishop and R.M. Lark

Abstract

In this comment on a recent article by Bishop and Lark, I show that standardized ordinary cokriging (Goovaerts, 1997, p. 232) is an inferior variant of customary cokriging. The authors' recommendation of standardized cokriging for the analysis of data, collected in landscape-scale field experiments, is therefore not justified. I further discuss the connection between customary ordinary cokriging and ordinary cokriging with a single unbiasedness constraint (Goovaerts, 1997, pp. 228–229). I argue that the latter method is not adequate for analysing experimental data because it excludes a priori that the treatments may have nonzero fixed effects that are constant across the study area. In contrast to Bishop and Lark, I recommend to use customary cokriging for the analysis of experimental data, gathered by landscape-scale experimentation.