Soil slaking assessment using image recognition.

We have developed a new methodology for the assessment of soil slaking under fast wetting conditions. We applied an image recognition algorithm to a set of digital images of soil aggregates immersed in water taken at regular time intervals. The kinetics of the slaking process was captured by measuring the projected aggregate’s area change over time. The methodology was tested in a dataset covering a great part of the agro-ecological variability of New South Wales (NSW), Australia. An empirical model which captures the rapid and slow slaking process was fitted to the data and three new slaking coefficients ( a , b and c ) were obtained and related to selected soil properties and land-use. The coefficient a , equivalent to the maximum slaking potential of the samples, was linearly related to exchangeable sodium, pH, clay percentage, calcium/magnesium and total carbon/nitrogen, and non-linearly related to total carbon. The coefficients b and c , equivalent with the initial slaking and the rate of change respectively, were found to be linearly related to nitrogen and total carbon. The coefficient a , was significantly lower in the natural sites reflecting a higher aggregate stability in those soils. The methodology is fast, inexpensive and simple; furthermore, it provides a new perspective in soil aggregate stability experiments, since it considers the slaking dynamics during the entire disaggregation process. [ABSTRACT FROM AUTHOR]