ESTIMATING SOIL TEXTURE FROM SAMPLE DENSITY

A number of laboratory methods for determining soil physical fractions have been proposed. The methods being cumbersome and time consuming c annot be adopted for routine analysis in soil testing laboratories. Th erefore, a simple method which could provide quantitative estimates of the soil size fractions is required. This study used soil sample dens ity (ratio of mass of soil in a standard scoop to its volume) to estim ate soil separates. Regression models were developed for predicting so il separates from sample density. Sand, silt, and silt+clay fractions were best described by second degree polynomials with coefficients of determination ranging from 0.75-0.80. Although the exponential functio n was statistically equivalent to the second degree polynomial for est imating clay, it was preferable as it yielded a better shape of the fi tted curve at sample densities < 1.2 g cm(-3). Validation of the model s with independent data showed that model-predicted values of various size fractions matched well with laboratory measurements with coeffici ents of correlation ranging between 0.95-0.99. The higher correlation coefficients between the measured and estimated clay, silt, and sand f ractions both for model development and their validation on independen t data suggest that the models could be used for estimating soil separ ates from sample density measurements.