A synoptic approach for crop loss assessment used to study wheat. III. Variables derived from soil properties and traditional soil classifications suitable for regression studies

Abstract

The combined information in the physical and chemical soil properties and in the traditional soil classification data has been split into three nearly uncorrelated groups of independent variables. One group comprises three variables and is concerned specifically with the traditional classification, a second group comprises seven variables and includes information specific to the soil properties, and a third group comprises one variable and includes the remainder of the information in the soil properties, i.e. that which is most strongly related to the traditional soil classification data. The new variables, which are substantially independent linear combinations of the original variables, form a convenient basis for the subsequent regression work in this study where the contribution of soil information to the variability of growth and yield of wheat in the study area is to be considered.