Ensemble pedotransfer functions to derive hydraulic properties for New Zealand soils

Abstract

Modelling water and solute transport through soil requires the characterisation of the soil hydraulic functions; however, determining these functions based on measurements is time-consuming and costly. Pedotransfer functions (PTFs), which make use of easily measurable soil properties to predict the hydraulic functions, have been proposed as an alternative to measurements. The better known and more widely used PTFs were developed in the USA or Europe, where large datasets exist. No specific PTFs have been published for New Zealand soils. To address this gap, we evaluated a range of published PTFs against an available dataset comprising a range of different soils from New Zealand and selected the best PTFs to construct an ensemble PTF (ePTF). Assessment (and adjustment when required) of published PTFs was done by comparing measurements and estimates of soil water content and the hydraulic conductivity at selected matric suction values. For each point, the best two or three PTFs were chosen to compose the ePTF, with correcting constants if needed. The outputs of the ePTF are the hydraulic properties at selected matric suctions, akin to obtaining measurements, thus allowing the fit of different equations as well as combining any available measurements. Testing of the ePTF showed promising performance, with reasonably accurate estimates of the water retention of an independent dataset. Root mean square error values averaged 0.06 m3 m–3 for various New Zealand soils, which is within the accuracy level of published PTF studies. The largest errors were found for soils with high clay content, for which the ePTF should be used with care. The performance of the ePTF for estimating soil hydraulic conductivity was not as reliable as for water content, exhibiting large scatter. Predictions of saturated hydraulic conductivity were of the same magnitude as the measurements, whereas the unsaturated values were generally under-predicted. The conductivity data available for this study were limited and highly variable. The estimates for hydraulic conductivity should therefore be used with much care, and future research should address measurements and analysis to improve the predictions. The ePTF was also used to parameterise the SWIM soil module for use in Agricultural Production Systems Simulator (APSIM) simulations. Comparisons of drainage predicted by APSIM against results from lysimeter experiments suggest that the use of the derived ePTF is suited for the estimation of soil parameters for use in modelling. The ePTF is not envisaged as a substitute for measurements but is a useful tool to complement datasets with limited amounts of measured data.

Saturated hydraulic conductivity model computed from bimodal water retention curves for a range of New Zealand soils

“…Cichota et al (2013) also reported that PTFs developed in Europe and the USA were not applicable to New Zealand. The reasons why these PTFs are not directly applicable to New Zealand require further investigation.…”

“…However, the predictions of K s were found to be too coarse for application to the wide range of soils within Smap. Therefore, Cichota et al (2013) tested published statistical PTFs developed in Europe and the USA to predict θ (h) and K(θ ) for a range of New Zealand soils. They combined the best two or three PTFs to construct ensemble PTFs.…”

“…Balland and Pollacco, 2008;Cichota et al, 2013), Pollacco et al (2013) developed a new physical model that predicts unimodal K s solely from hydraulic parameters derived from the Kosugi (1996) θ (h). The K s model is derived by combining the Hagen-Poiseuille and Darcy laws (Sutera and Skalak, 1993) and by incorporating three semi-empirical tortuosity parameters.…”

“…Therefore, these curves are generally retrieved from pedotransfer functions (PTFs), which are statistical relationships that generate lower-precision estimates of physical properties of interest based on many rapid and inexpensive measurements (e.g. Balland and Pollacco, 2008;Pollacco, 2008;Anderson and Bouma, 1973;Webb, 2003;Cichota et al, 2013).…”

Development and Uncertainty Assessment of Pedotransfer Functions for Predicting Water Contents at Specific Pressure Heads

“…Similar functions were developed and tested later by Minasny et al (1999), Tomasella et al (2000), and Børgesen et al (2008) for soils from New Zealand, Brazil, and Denmark, respectively. Further developments beyond the classical regression analysis include techniques like artificial neural networks (Schaap and Leij, 1998a, 1998b; Minasny and McBratney, 2002; Merdun et al, 2006; Baker and Ellison, 2008; Campos de Oliveira et al, 2017; D’Emilio et al, 2018), pattern recognition based methods including support vector machine (Nemes et al, 2006; Lamorski et al, 2008; Khlosi et al, 2016), Gaussian process regression (Kotlar et al, 2019b), and the ensemble approach (Baker and Ellison, 2008; Cichota et al, 2013; Liao et al, 2015), all of which contributed to the improvement of PTF performance in predicting soil hydraulic properties.…”