Evaluation of a Model for Predicting Compressibility of Forest Soils

Abstract

AbstractSoil compressibility is an important soil physical property in the prediction of compaction behavior. The compression index, C, as a measure of compressibility, was determined for a wide range of South African forest soils and compared with existing models that utilize clay content as the independent variable. These models were tested using a model performance technique that distinguished between systematic (bias) and nonsystematic (random) errors. The maximum compression index, Cmax, was adequately predicted for all soil orders in this study and possessed a relatively low systematic error. The average compression index, Cmod, which was calculated in this study as the average of C values across a wide range of water contents, was generally underpredicted and showed a slightly higher systematic error than with Cmax. It is concluded that the slightly better performance of the models in predicting Cmax was due to the models being developed on moist to wet soils (Ψp > −100 kPa) when compression is at a maximum. Since compaction can also occur under moist to dry conditions, our results indicate that modelling compaction behavior for soils that occur in regions with a pronounced dry season will require compression indices developed across a wider range of water contents than is currently the case.