A simplified surface lithology class map for Australia
Jonathan Gray1 and John Wilford2
1 NSW Department of Environment and Planning
2 Geoscience Australia
20 April 2022
A digital map of lithology classes has been prepared over Australia, primarily for use as a variable in digital soil mapping and related environmental modelling applications. The map was based on Australian geological 1:1 million scale mapping, with unconsolidated Cainozoic materials being based on 1:2 million scale soil mapping. It is available in both polygonal and raster format.
Lithology was grouped into 11 classes, with 8 classes based on siliceous character (i.e., ultra-mafic to extremely siliceous), plus 3 other non alumino-silicate materials (calcareous, sesquioxide and organic). Grouping into these 11 classes enhances the performance of this variable in digital soil modelling and mapping applications relative to using a larger number of less well-defined geological units or other parent material classes.
Preliminary testing suggests moderate performance within DSM applications that slightly improves on using geophysical variables alone. Although the raster resolution is 100 m, the broad scale of the underlying source mapping (1: 1-2 million) means the product is only reliable at national or broad regional scales. Lithology data collected in the field or from finer scale maps, and similarly classified into the 11 classes, should be applied in preference to this product.
Parent material is a key factor of soil formation. It is one of the five factors (p) in the fundamental soil equation of Jenny (1941) (cl,o,r,p,t) and one of seven factors in the s,c,o,r,p,a,n model of McBratney et al. (2003). The chemical composition of the parent material influences the chemistry of the resulting soils.
In DSM, parent material may be represented by lithology derived from available geological and soil maps in addition to remotely derived geophysical data (eg, gamma radiometrics or hyper spectral data). The former source is typically categorical and polygonal in format, whilst the latter is typically raster based and continuous in format, with fine resolution down to 100 m or less.
Gray et al. (2016) demonstrated the strong influence of parent material variables over a suite of soil properties in New South Wales, particularly lithology. For example, in maps of cation exchange capacity (CEC) over the 5-15 cm interval, Lin’s concordance (LCCC) rose from 0.44 with no parent material covariates, to 0.50 with geophysical covariates to 0.60 with lithology variables alone, a 36% increase compared to no parent material covariates and a 20% increase relative to geophysical covariates alone. It is evident that strong relationships are present between soil properties and lithology, but relationships between soil properties and geophysical data are often less clear.
Application of lithology data from geology maps is frequently hindered by the large number of different geological units involved, all with slightly differing descriptions. The use of a dominant lithology field that is often provided may be useful in reducing the number of classes, but these descriptors are frequently not sufficiently detailed or rigorous, and may not adequately distinguish between different units, for example, grouping diorites with granites, or feldspathic sandstones with quartz sandstones, or all unconsolidated material (e.g., clays and sands) together.
The product presented here attempts to group the parent materials of the Australian continent into just 11 broad classes, facilitating its use for broad scale modelling applications across Australia.
The classification into 11 classes closely followed the classification scheme presented in Gray et al. (2016), but omitting the evaporite class, as shown in Table 1.
The primary source of geology was the 1:1 M scale geological map of Australia (Geoscience Australia 2022). The original 245 000 geological units were initially sorted by “LithGroup1” (dominant lithology) then “LithGroup2” (sub-dominant lithology), then by lithological description. Each geological unit was examined and allocated into one of the eleven classes. The dominant and subdominant lithologies alone were not sufficient to reliably group each unit into one of the eleven classes; the detailed descriptions were also required.
In many cases, there was a degree of subjectivity when rock types of differing chemistry were included in the same geological unit, for example, if the unit comprised interbeds of quartz sandstone and shale, or granite with diorite. In such cases, an approximate weighted average of the lithology classes was applied.
The descriptions of Cainozoic unconsolidated material in the geology map were found to be insufficient to reliably group the material (e.g., being mixtures of sand, silt and clay with no clear estimate of relative proportions and variable over broad regions). In these cases, available soil mapping was preferred. The 1:2 M Atlas of Australian Soils (Northcote et al. 1960-1968) was used. The different soil types were likewise grouped into 11 classes as shown in Table 2, based on their typical composition, based on sources such as Stace et al. 1968).
3. The final products
The estimated lithological class map is presented as a raster (geotiff) file with 0.001 degree (approx. 100 m) resolution and a polygonal shapefile. The raster provides a code number (1-14), which can be linked to the associated lithological class name with the provided layer file, as shown in Figure 1 below.
Figure 1: Estimated lithology classes over Australia
A shapefile with polygonal data containing an attribute table with further specific detail on the geologic and/or soil map unit was also prepared. Figure 2 provides a screenshot of the layer over north-east Australia, with an example identification of a specific unit. Where the surface material is hard rock, there is a “G” prefix and the details relate to the geological map descriptions; where material is unconsolidated regolith, there is an “S” prefix and the details given relate to the Soil Atlas descriptions. In some cases, both soil and hard rock descriptions are provided. In addition to Lith_class and code (“New_code1”), an estimate of silica % is also provided, except for the non-siliceous carbonate, organic or sesquioxide materials.
Figure 2: Screenshot of lithology class shapefile with example attribute data over NE Australia
4. Preliminary testing of the lithology layer
Preliminary testing of the reliability of the estimated lithological class was carried out by comparing the estimate with observed data from 13 364 points of a subset of the Ozchem dataset (Geoscience Australia, unpublished ). This subset comprised bedrock and regolith samples, each with the observed silica concentration (%) and rock or regolith material name.
The results are presented in Figure 3 below and indicate 38% of estimates were equivalent to observed values and 73% were within 1 class unit of observed values. Approximately 17% of estimates were more than 2 classes different to the observed values, thus generally unacceptably different, with most of these being in the arid interior with unconsolidated regolith overlying bedrock.
Figure 3: Difference between observed and estimated lithology class
The lithology layer has been tested in models for digital soil mapping over NSW and eastern Australian.
Variable importance plots from random forest models demonstrate the lithology layer being one of the most influential of all variables used. For sum-of-bases over NSW, it was the most influential (Figure 4a) while for SOC over all eastern states, it was the 3rd most influential (behind the climatic variables, Figure 4b).
Figure 4: Variable importance plots from random forest models
Comparisons of validation statistics using the above random forest modelling with and without the lithology layer are presented in Table 3a & b. An improvement in model performance, albeit only slight, is demonstrated with the addition of the lithology layer, compared to using only geophysical parent material (PM) variables such as gamma radiometrics (Geoscience Australia 2023), NIR spectroscopic clay components (Viscarra Rossel 2011) and weathering index (Wilford 2012). It was noted that using multiple linear regression modelling demonstrated a more significant improvement in performance with the use of lithology than evident from using random forest models.
Table 3: Validation statistics for SOC and sum-of-bases models, NSW eastern Australia and NSW
The Australia-wide lithology layer as presented here can provide a useful additional variable in digital soil mapping, to complement other purely geophysical parent material variables. The use of just 11 meaningful lithology classes makes it easier to apply than currently available broad scale geological maps. Preliminary testing suggests moderate performance that slightly improves on using geophysical variables alone. It can be useful in helping to interpret the role of geology in soil maps compared to geophysical data that is difficult to relate to on-ground conditions, for example, the meaning and significance of high radiometric K, U or Th is not readily apparent.
It is reiterated that the best source of geological data for a specific point (as used in developing models) is derived from field survey or fine scale geological mapping. Those sources should be relied upon for point data ahead of the broad scale lithological map presented here.
Charman, P.E.V., Murphy, B.W. Soils: Their Properties and Management. A Soil Conservation Handbook for NSW'. Third Edition. Oxford University Press, Melbourne.
Duff, D., 1993. Holmes’ Principles of Physical Geology, London - New York, Chapman and Hall.
Geoscience Australia 2022. Geology of Australia and Australian Antarctic Territory. Geoscience Australia, Canberra. https://ecat.ga.gov.au/geonetwork/srv/eng/catalog.search
Geoscience Australia 2023. AWAGS Radiometrics and Magnetics Project, https://www.ga.gov.au/about/projects/resources/awags
Gray JM, Bishop TFA, Wilford JR, 2016. Lithology and soil relationships for soil modelling and mapping, Catena 147, 429–440, https://doi.org/10.1016/j.catena.2016.07.045
Gray, J.M., Humphreys, G.S., Deckers, J.A, 2011. Distribution patterns of World Reference Base soil groups relative to soil forming factors. Geoderma 163, 373-383.
Gray, J.M., Murphy, B.M., 1999. Parent Material and Soils: a Guide to the Influence of Parent Material on Soil Distribution in Eastern Australia. Technical Report No. 45, NSW Department of Land and Water Conservation, Sydney.
Isbell RF and NCST 2021, The Australian Soil Classification, Third Edition, with National Committee on Soil and Terrain, CSIRO Publishing, Melbourne.
IUSS Working Group WRB. 2022. World Reference Base for Soil Resources. International soil classification system for naming soils and creating legends for soil maps. 4th edition. International Union of Soil Sciences (IUSS), Vienna, Austria.
Joplin, G.A., 1965. Chemical Analysis of Australian Rocks. Part II: Sedimentary Rocks. Bureau of Mineral Resources, Geology and Geophysics, Bulletin No. 78. Canberra.
Mason, B., 1966. Principles of Geochemistry. 3rd edition, John Wiley and Sons Inc.; Toppan Company Ltd, Tokyo.
McBratney AB, Mendonca Santos ML, Minasny B (2003) On digital soil mapping. Geoderma 117, 3-52. https://doi.org/10.1016/S0016-7061(03)00223-4
Northcote, K. H. with Beckmann, G. G., Bettenay, E., Churchward, H. M., Van Dijk, D. C., Dimmock, G. M., Hubble, G. D., Isbell, R. F., McArthur, W. M., Murtha, G. G., Nicolls, K. D., Paton, T. R., Thompson, C. H., Webb, A. A. and Wright, M. J. (1960-1968). Atlas of Australian Soils, Sheets 1 to 10. With explanatory data (CSIRO Aust. and Melbourne University Press: Melbourne).
Pettijohn, F.J., 1963. Chemical composition of sandstone – excluding carbonate and volcanic sands., in: Fleischer M. (Ed.), ‘Data on Geochemistry’, 6th edition, Geological Survey Professional Paper 4440_S, United States Department of Interior, US Government Printing Office, Washington.
Stace, H.C.T., Hubble, G.D., Brewer, R., Northcote, K.H., Sleeman, J.R., Mulcahy, M.J., and Hallsworth, E.G. 1968, A Handbook of Australian Soils, CSIRO and International Society of Soil Science, Rellim Technical Publications, Glenside, South Australia.
Viscarra Rossel, R.A., 2011. Fine‐resolution multiscale mapping of clay minerals in Australian soils measured with near infrared spectra. Journal of Geophysical Research 116, F04023, 15p
Wilford, J., 2012. A weathering intensity index for the Australian continent using airborne gamma-ray spectrometry and digital terrain analysis. Geoderma 183, 124-142