Estimating Total Acidity From Soil Properties Using Linear Models

Currently, there is no national or comprehensive model for estimating total acidity (extractable acidity) that uses basic soil properties obtained in soil survey. The aim of this study was to develop total acidity prediction models from readily available soil properties. Measured data in the National Soil Survey Characterization Database were used to develop the total acidity prediction equations. The data set was stratified primarily by taxonomic order. Strata were also separated by texture modifiers of hydrous, medial, and ashy, and by different organic layers. Prediction equations were developed for each strata (or data set) using general linear model procedures. Organic C, pH (in H₂O), and cation exchange capacity (CEC) were the most highly correlated and most important variables in predicting total acidity. Two sets of total acidity prediction equations were developed. One set uses CEC as a predictor variable, and the second set provides total acidity estimates for low pH soils (pH <5.5) where CEC is not available. The two sets of equations allow total acidity to be predicted for all soils in soil survey. The coefficient-of-determination (R²) ranged from 0.55 to 0.93 for prediction equations using CEC as a variable and from 0.44 to 0.86 for prediction of total acidity for low pH soils, where CEC is not used as a predictor variable. Validation of the equations indicated that these models can be useful in estimating total acidity with decreasing reliability at lower pH (when CEC is not used). The most useful equations are those with R² > 0.60, which can be useful in estimating total acidity for soil survey when there are no measured data available.