The aim of this research was to enable erodibility values for hillslope-scale
erosion prediction models to be determined from easily measured media
properties. Simulated rainfall and overland flow experiments were carried out
on 34 soils and overburdens from 15 Queensland open-cut coal mines at The
University of Queensland Erosion Processes Laboratory. Properties of the 34
media determined included aggregate stability, Atterberg limits, bulk density,
cation exchange capacity, dispersion ratios, electrical conductivity,
exchangeable sodium percentage, organic carbon content, pH, texture, and water
content at field capacity and wilting point. Correlation and stepwise multiple
regression procedures were used to determine those media properties that could
best be used to predict rill and interill erodibility. Correlations between
media properties and sediment delivery at each of 5, 10, 15, 20, and
30% slope revealed that different media properties were correlated with
erosion rates at different slopes. A media property could show a strong
correlation with erodibility at 30% slope, and a low correlation at
5% slope. Splitting the data set into soils only, and overburdens only,
showed that properties that were positively correlated with erosion rates for
one group could be negatively correlated for the other group. Therefore, in
this study, erodibility could not be explicitly linked to one set of media
properties for all medium types and erosive conditions. It was concluded that
a single regression equation could not be used to predict erodibility under
all conditions. Instead, 4 equations were developed to predict rill and
interill erodibility, for soils and overburdens separately. The need for
separate regression equations was attributed to the presence of different
erosive sub-processes for specific combinations of medium type and slope
gradient.