Prediction of three key hydraulic properties in a soil survey of a small forested catchment

Direct measurement of psi(theta) and K(theta) relationships at all observation sites in soil survey is not feasible. Three key hydraulic properties-water content at field capacity (theta(-5 kPa)), water content at wilting point (theta(-1.5 MPa)), and saturated hydraulic conductivity (K(s))-can be used to derive K(theta) and psi(theta) when combined with bulk density. These properties were measured in 'calibration' horizons in a soil survey in Yambulla State Forest in south-east New South Wales. Pedotransfer functions (PTFs) for predicting theta(-5 kPa), theta(-1.5 MPa), and K(s) from the physical and morphologic soil attributes are presented and evaluated here. Models for predicting theta(-5 kPa) and theta(-1.5 MPa) relied on per cent clay. An R(2) of 0.64 (for theta(-5 kPa)) to 0.67 (for theta(-1.5 MPa)) was obtained for linear regressions using only morphologic explanatory variables. An R(2) of 0.73 (for theta(-5 kPa)) to 0.90 (for theta(-1.5 MPa)) was obtained if laboratory-measured clay content was included as an explanatory variable. K(s) was measured in situ using well permeameters, and used for developing PTFs. Large cores were taken from a small subsample of horizons and measurements of K(s), K(-0.1 kPa), K(-0.2 kPa), and K(-0.5 kPa) were made in the laboratory. K(s) measurements from well permeameters were similar to K(-0.5 kPa) from laboratory measurements. Regression and tree models were used to predict K(s). The linear regression had an R(2) of 0.55, while the tree models accounted for approximately 40% reduction in deviance. Bulk density was the most useful predictor in all K(s) models. The inclusion of per cent rock fragments, bulk density, and estimated percentage clay as useful explanatory variables demonstrated the utility of functional descriptors not routinely measured in soil survey. The models are empirical and were locally calibrated for use in a soil survey. They may be applicable in target domains similar to the source domain (i.e. coarse-grained adamellite soils in similar climatic regimes).