LAUSR

Abstract Plant available water capacity (PAWC), defined as the amount of water held in the soil between the drained upper limit (DUL) and crop lower limit (CLL), is a major determinant for crop yield in dryland cropping areas. However, measured soil profile data quantifying the physical soil parameters that determine PAWC are limited. Choosing a ‘representative’ soil profile in order to parameterise a deterministic model is often subjective and may not be appropriate, leading to biased conclusions. To help support these deterministic crop modelling studies, we analysed soil profile data in Australia available in the APSoil database; a database of field and laboratory measured physical soil properties used to parameterise the Agricultural Production Systems sIMulator (APSIM) model. Specifically, we developed typical soil pattern profiles utilizing wheat for CLL using statistical shape analysis and cluster analysis. Our method allows modellers to reduce the number of soil profile parameter choices using a statistically sound basis. We implement our method by grouping the APSoil soil profiles into five clusters, whereby each cluster is represented by a more general soil profile. The representative soil profiles reflect impact of the soil texture on PAWC. The five clusters and representative soil profiles can then be used in future modelling studies to investigate the impact of soil variation on systems performance. In turn, soil scientists will have a plausible and quantifiable collection of soil profiles to choose from for future studies.