RELATIONSHIPS BETWEEN CONVENTIONAL FIELD INFORMATION AND SOME SOIL PROPERTIES MEASURED IN THE LABORATORY

Soil variables measured in the laboratory were correlated with field information obtained from a routine soil survey of the Peace River region of British Columbia, Canada. The laboratory variables included bulk density, carbon content, and gravimetric water contents at saturation, field capacity and wilting point. The field information included colour, texture, structure, horizon type and effervescence. The modelling was performed in a two-phase statistical procedure. In the first phase, categories within each variable of field information with means of laboratory variables not significantly different at the 5% level, were integrated into one category. In the next phase, all field variables with distinct means were transformed into "dummy variables" assigned with the numerical values of - 1, 0 or 1. Using a step-down stepwise regression technique for model analysis, all field variables which correlated significantly with each laboratory variable were identified. A multiple regression equation was ultimately proposed for estimation of each laboratory variable. The results indicated that carbon content and water content at field capacity could be accurately estimated (R2 > 0.70) using field variables. Other laboratory variables were only moderately predicted for the Peace River soils.