Testing Logselfsimilarity of Soil Particle Size Distribution: Simulation with Minimum Inputs

Abstract

Particle size distribution (PSD) greatly influences other soil physical properties. A detailed textural analysis is time-consuming and expensive. Soil texture is commonly reported in terms of mass percentages of a small number of size fractions (typically, clay, silt and sand). A method to simulate the PSD from such a poor description or even from the poorest description, consisting in the mass percentages of only two soil size fractions, would be extremly useful for prediction purposes. The goal of this paper is to simulate soil PSDs from the minimum number of inputs, i.e., two and three textural fraction contents, by using a logselfsimilar model and an iterated function system constructed with these data. High quality data on 171 soils are used. Additionally, the characterization of soil texture by entropy-based parameters provided by the model is tested. Results indicate that the logselfsimilar model may be a useful tool to simulate PSD for the construction of pedotransfer functions related to other soil properties when textural information is limited to moderate textural data.