A comparison of soil survey methods in relation to catchment hydrology

Abstract

Eight primary catchments within the Western Australian wheatbelt were surveyed in detail to examine the abilities of conventional soil classification and geostatistical analysis to provide detailed information of soil spatial variation for catchment-scale hydrologic modelling. Nine soil physical properties were measured. The results illustrate potential diculties with both methods. Classification by using the Factual Key was unable to describe the major component of soil property variation. The relative variance accounted for by soil classes was usually <10%. Only the yellow duplex soils appeared distinct from other soil classes. Potential diculties with geostatistical analysis also arose because of fluctuations in the variogram models. Contrasts occurred between variograms for the same property over different catchments and for different properties over the same catchment. Within the areas studied, nugget and linear (unbounded) variogram models were more common than spherical or exponential models. It is proposed that the surveyor would have to select a survey method on the basis of prior knowledge about which model of variation is more likely to be successful for the scale and location of survey.