Evaluation of different procedures to interpolate particle size distribution in black soils

Abstract

The particle-size distribution (PSD) model can be used to describe a full range of soil PSD from sparse experimental data. To determine the best model in representing PSDs of black soils in northeast China, five popular PSD procedures were examined for 78 tested soils. The coefficient of determination (R2), model efficiency (E), root mean square error (RMSE), mean absolute percentage error (MAPE) and absolute relative error (|RE|) were used to measure the goodness-fit of the procedures. Three original PSD data classes (diameters 2, 20, 200 μm) were provided to interpolate the mass fraction for a particle diameter < 50 μm. Our results indicate that the log-normal interpolation and modified logistic growth model were best for the majority of soils studied, and the interpolated results were satisfactory. Special attention was given to the effect of texture on model performance, and performance of models was best when estimating loam, while clay content does not affect performance PSD models.