Embedded Files
Ross Searle, P.D.S.N Somarathna & Brendan Malone
Along with being a key environmental driver, soil water content is a critical piece of information needed to make optimal management decisions in agricultural production systems. Commonly primary producers rely on rainfall records, knowledge of current management and expert knowledge based on experience to estimate soil water contents. Whilst these estimates are often “good enough” it is difficult just using these heuristics to know with enough detail, for critical management decisions, how much water is available in the soil profile and where that water is located in the profile.
We employed standard Digital Soil Modelling methods utilising publicly available soil observation data and publicly available environmental covariate data in an environmental correlation approach using machine learning to map the soil properties of volumetric (mm/mm) drained upper limit (one third bar, DUL), soil lower limit (15 bar, LL15) and total Available Water Capacity (AWC) across the entire continent at 6 standard depths at 90m pixel resolution.
We used pedotransfer functions for estimating DUL and LL15 from readily available soil attribute data have recently been developed by Somarathna and Searle (in press) using data from the National Soil Site Collation (NSSC). ). The National Soil Site Collation (NSSC) database contains 1190 laboratory measurements of DUL and LL15 that were used to develop pedotransfer functions using multilinear regression (Fig 2) against the independent variables of clay, sand & silt percentages, soil organic carbon and bulk density.
The derived MLR equations for the estimating volumetric DUL (Eqn 1) and LL15 (Eqn 2) are given below. The observed vs predicted model fits with concordance statistics are shown in the figure below. Measured soil property data was obtained using the TERN SoilDataFederator
Model validations
Model validations
The table below summarises the external validation outcomes for both LL15, DUL and AWC across the standard depth intervals. In general model fits are better at the surface and decline with depth. LL15 predictions tend to be slightly better than those of DUL at the same depths This is most likely associated with a decrease in available data at depth. These model fits are quite strong in a DSM context. Figure 7 shows the scatter plots of observed vs. predicted values of LL15 (a) and DUL (b) across the standard depth intervals
The spatial modelling of DUL and LL15 is done at six standard depth intervals conforming to the GlobalSoilMap Specifications ie 0-5cm, 5-15cm, 15-30cm, 30-60cm, 60-100cm and 100-200cm. To facilitate modelling at these standard depths the observed data set depths were harmonised to these depths using a mass preserving spline method. A total of 20545 soil profiles were splined in this way and used as inputs to the spatial modelling.
We utilised the publicly available Terrestrial Ecosystem Research Network (TERN) raster covariate stack. It is comprised of 154 individual raster data layers. The covariate stack was used as the independent variable data for the predictions across all grid cells and at each depth. Environmental correlation using the Cubist machine learning algorithm was used to develop relationships between the soil property value estimates derived from the pedotransfer functions and the values of the covariates.
Fifty bootstrapped model realisations, using the Cubist machine learning algorithm (Quinlan, 1992), were generated, and used to predicted DUL and L15 values (mean of the bootstrap realisations) and estimate upper and lower confidence intervals (5% and 95%) across the entire continent
The Available Water Capacity values were calculated by subtracting the L15 values of each layer from the DUL values of each layer and the upper and lower confidence intervals were estimated by combining the variances of the upper and lower confidence intervals of L15 and DUL.
To estimate the Total Available Volumetric Water Capacity (mm) to 1 and 2 metres we summed all the AWC layer values converted to mm of water to the estimated soil depth (Australia-wide 3D digital soil property maps - Depth of Soil (3 arc second resolution - Version 2) or the designated depth of the product - which ever was shallowest.
Maps of LL15, DUL and AWC at the six depth intervals were generated and are presented below, along with maps of Total Available Water Capacity in mm of water for 0 to 100 cm and 0-200.
Predicted Lower Limit 15 Bar at the six standard depths.
Predicted Drained Upper Limit 1/3 Bar at the six standard depths
Predicted AWC at the six standard depths.
Total Available Water Capacity in mm of water for (a) 0 to 100 cm and (b) 0-200 cm or shallower depending on estimated soil depth.
Observation data - https://esoil.io/TERNLandscapes/Public/Pages/SoilDataFederator/SoilDataFederator.html
Covariate rasters - https://esoil.io/TERNLandscapes/Public/Pages/COGs
Page updated
Google Sites
Report abuse