Chapter 32 Digital Mapping of Soil Attributes for Regional and Catchment Modelling, using Ancillary Covariates, Statistical and Geostatistical Techniques

Abstract

The application of statistical techniques to spatially predicting soil attributes from ancillary variables evolved from Jenny's factors of soil formation, termed “Climate, Organisms, Relief, Parent material and Time” or “ corpt ”. The corpt approach was recently extended to include other soil attributes ( s as surrogates) and space ( n ), and thus it is termed scorpan with time factor t in corpt replaced as age ( a) . The main objectives of this chapter are to collate and integrate various land feature digital layers to the same resolution and coordinate system, and to develop spatial prediction models based on scorpan and scorpan -kriging methods, for predicting selected soil attributes. Existing analogue maps were first digitised and transformed into digital maps. These, along with other existing digital information, were reprojected in the same coordinate system, the Geocentric Datum of Australia (1994) and Map Grid of Australia (1994), namely GDA-94 and MGA-94. Further these digital map layers, along with Landsat bands and digital terrain attributes, were used to predict soil attributes and thus producing different soil attribute maps for a number of soil depths. The spatial prediction methods used were scorpan methods, such as multiple linear regressions (MLR) and scorpan -kriging (SK), which combines simple kriging with MLR. While MLR was good enough model to predict a number of soil attributes, SK was more appropriate for electrical conductivity for two layers: 0–10 and 70–80 cm layers and was equally good, if not slightly better than the scorpan technique of MLR. However, the results of scorpan- kriging exhibit more detailed variation across the extent of the study area compared with kriging or MLR. The application of generalised linear and generalised additive models did not improve the prediction accuracy. Two of the soil attributes – clay content and electrical conductivity – for both the topsoil (0–10-cm depth) and subsoil (70-80-cm depth) are illustrated here. Both soil attributes have significant influence on the hydrological processes shaping the landscape. Finally, the digital land feature maps, including those of soil and digital terrain attributes, are displayed as geographical information systems (GIS) layers, which could potentially be useful for various environmental and catchment modelling.