Comment on the article on ‘Within‐field wheat yield prediction from IKONOS data, a new matrix approach’ by E. A. Enclona, P. S. Thenkabail, D. Celis and J. Diekmann, International Journal of Remote Sensing, 25, 377–388 (2004)

Abstract

Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tres20 Comment on the article on 'Within-field wheat yield prediction from IKONOS data, a new matrix approach' by E. A. Enclona, P. S.This article comments on a method of within-field yield prediction from IKONOS data published in this Journal (International Journal of Remote Sensing, 25, 377-388). The authors propose what they call a new matrix approach, which solves a system of linear equations relating actual yield to expected yield in four classes defined based on spectral characteristics and the relative areas for these classes as measured by remote sensing. The authors use five observations to estimate four unknown parameters, and they propose that generally p + 1 observations could be used to estimate p parameters. This article shows that this approach will lead to very unreliable predictions of class yield. The problem is shown to be a standard multiple regression problem. It is argued that sample size must be considerably larger than the number of classes in order to obtain reliable yield estimates for the classes. Due to the very nature of the regressor variables, the relative areas of the classes, multicollinearity is expected to be a severe problem. Therefore, it may be useful to reduce the number of regressor variables using standard model selection procedures.