Soil classification and survey studies at Ginninderra

doi:10.1071/SR9760001

Soil classification and survey studies at Ginninderra

R Webster and BE Butler

Australian Journal of Soil Research 14(1) 1 - 24
Published: 1976

Abstract

Studies were made of the interrelations of soil properties, their spatial variation, and the relations between sampling sites at Ginninderra Experiment Station near Canberra. Soil morphology was recorded at 0-10 cm, 15-25 cm (usually A2 horizon), 41-51 cm (usually B horizon) and 66-76 cm, and general profile features were noted. Samples from the 0-10 cm and 41-51 cm were analysed chemically, and bulk density and water held at 0.1 bar determined on 0-7.5 cm cores. Simple correlations between morphological and other properties of the topsoil were almost all negligible. Correlations in the B horizon were somewhat stronger, though few correlation coefficients exceeded 0.5 in absolute value. Ordination by principal coordinate analysis showed sample sites to be evenly spread in a single roughly spherical cluster in the three main dimensions of property space. Four classifications, two systematic, one a map, and the fourth a numerical classification, were unable to predict values of soil properties sufficiently well to be useful; the properties must be measured where they are of interest by a purpose-designed sampling scheme. Soluble potassium varies little between points 5 m apart, but very considerably over distances 56-180 m. Two-thirds of the variation in soil phosphorus at Ginninderra occur within 5 m and the remainder at more than 180 m. Variation in pH increases fairly steadily up to 56 m, and the rest at more than 180 m. Most variation in bulk density and water content occurs within 18 m, though the total variation is small. Variation in morphology occurs mainly within 5 m and at more than 180 m. Reasonable maps can be made of phosphorus and morphology by pooling data from samples from 5 m diameter circles, or by bulking samples from them, spaced on a grid of approximately 200 m mesh. For potassium a grid mesh of no more than 56 m is needed to map the variation, though pooling data from 5 m circles would give a small improvement.