Variability of soil erodibility: its relationship to topography and soil properties in cultivated landscape

Abstract

The hypothesis of this thesis is that soil erodibility varies systematically with landscape topography. This hypothesis was tested by using a portable rainfall simulator to measure the relative soil erodibility at five slope positions along eleven slope transects on fallow agricultural land in hummocky terrain. Soil erodibility variations among the five slope units were examined for significant differences. The slope units were found to be ranked in descending order by mean soil erodibility as follows: shoulder, midslope, footslope, summit, toeslope. This systematic pattern of erodibility was evident even though topographic characteristics and soil properties varied substantially between the slope transects sampled within the small study area. Differences in erodibility were attributed to soil property differences along the slopes. Correlation and multiple regression indicated that soil properties had a stronger relationship to soil erodibility than did topographic variables. Particle size distribution, bulk density and organic matter had the most significant effect on erodibility variations. Aggregation measures were not found to be good indicators of soil erodibility. Topography was thought to influence erodibility indirectly through its effect on the distribution of soil properties along the slope transects based on the higher correlations between topographic variables and many soil properties. Higher correlations between erodibility and soil and topographic variables were found when analyzed by slope position, especially on the midslope. This indicates that soil properties were more uniform when grouped by slope position than when considered as a single, large data set. This study suggested that the use of a catena approach to examine soil erodibility is an important consideration when studying soil erosion processes in agricultural landscapes.