
Appendix 6
History of the development of the Australian Soil Classification
Activities that led to the Australian Soil Classification (ASC) commenced in 1981, when the Soil and Land Resources Committee (SLRC, then a sub-committee of the Standing Committee on Soil Conservation-SCSC) recommended the formation of a working party to look into the need for and the options for improving soil classification in Australia. In 1982 a questionnaire on the subject was sent to all members of the Australian Soil Science Society, the results of which were published (Isbell 1984). The working party (Ray Isbell, Pat Walker, David Chittleborough, Robert van de Graaff, Ron McDonald) recommended to the SCSC (via the SLRC) in 1984 that a soil classification committee be established under the auspices of SLRC to formulate a proposal for the establishment of a new or revised Australian soil classification. The working party also listed various options for this task, and provided a number of guiding principles.
The soil classification committee was formally endorsed by the SCSC early in 1985, with the following membership: Ray Isbell (Convener), David Chittleborough (SA), Alex McBratney (QLD), Ron McDonald (QLD), Brian Murphy (NSW). Ian Sargeant (VIC) joined early in 1986.
The committee first met in August 1985 in Brisbane - Ken Day (NT) also attended. This meeting endorsed with some amendment the guiding principles of the earlier working party, and examined the various options available for a new or revised classification, particularly in the light of the replies of the earlier questionnaire.
The various options considered were:
- Revision of the existing Stace et al. (1968) great soil group scheme. This was not considered practical but the scheme could be partly used as a basis for the preferred option.
- Revision of the Factual Key (Northcote 1979). This was not practical given the structure of the classification. Also, it cannot strictly be considered as a general purpose scheme given the limited nature of the attributes used. However, appropriate features of the system could be incorporated into any new classification scheme.
- Adoption of an overseas scheme, for example Soil Taxonomy (Soil Survey Staff 1975) or FAO-Unesco (1990). The data base on which these schemes were constructed related mostly to northern hemisphere temperate zone soils, therefore, it could not be expected that these would be the most appropriate for Australian soils. Experience has shown this to be true.
- Adaptation of an overseas scheme to Australian needs and conditions. This was thought to be quite impractical and would also lead to confusion.
- Development of a computer-based numerical system. Although some experiments have been conducted, no such scheme had yet been developed on a national basis anywhere in the world. Although techniques were becoming available, the lack of standardised data is and will continue to be a problem for the foreseeable future.
During and following the 1985 committee meeting, attempts were made to establish likely diagnostic horizons, and existing classes of Australian soils - for example, Stace et al. (1968) great soil groups and some Factual Key classes - were grouped into provisional new classes at various hierarchical levels. A meeting in July 1986 devoted particular attention to the question of creating classes using numerical methods. Subsequent exercises using the computer-based fuzzy set techniques developed by Alex McBratney were tried. Although this method had merit, an insurmountable problem at the time was the lack of an adequate representative data set.
The selected option for a new Australian classification system was for a multi-categoric scheme with classes defined on the basis of diagnostic features and their arrangement in vertical sequence as seen in an exposed soil profile, rather than geographic attributes were to be used. In the new scheme, classes are based on real soil bodies, they are mutually exclusive, and the allocation of 'new' or 'unknown' individuals to the classes is by means of a key.
In March 1987, a preliminary version of the classification was sent to 25 pedologists around Australia for comment. The many useful replies were considered by the Committee at a meeting in Sydney in July 1987. Due to lack of any funding arrangements, no formal meetings of the Soil Classification Committee took place until it was reconstituted through the Working Group on Land Resource Assessment and the Australian Collaborative Land Evaluation Program in the early 1990s.
In late 1989, a 'First Approximation' of the scheme was issued as an unpublished working document (CSIRO Division of Soils Technical Memorandum 32/1989). This was widely distributed to some 200 people throughout Australia - many as a result of requests. The period from late 1989 to late 1991 was devoted to extensive testing of the scheme, both in the field and by checking relevant publications, and to a lesser extent by interrogating the CSIRO Division of Soils data base.
A 'Second Approximation' issued in January 1992 was a very much expanded version of the earlier one. Although the number of Orders remained the same, one was dropped (Melanosols) and one added (Dermosols). The main reason for the omission was that the diagnostic surface horizon of Melanosols is too easily lost by erosion or modified by human action - a problem similarly encountered in the Mollisols of Soil Taxonomy. The other major change was the narrower definition of Ferrosols as soils with high iron contents.
The introduction of Dermosols catered for similar structured soils that lack high iron contents.
In August 1992, the Australian Soil Conservation Council formally endorsed the new classification and recommended its adoption by all States and Territories and its use in all future federally funded land resource inventory and research programs. During 1992-93, a National Landcare Program grant enabled extensive field travel around Australia, and provided for an assistant to carry out extensive testing of the classification via published data and unpublished material in data bases.
The 'Third Approximation' (Isbell 1993) followed extensive testing during 1992, both in the field and by checking relevant publications and, in particular, the comprehensive Queensland Department of Primary Industries soil profile data base. Over 300 copies of this version were distributed to individuals and organisations, as well as copies to the approximately 70 people actively engaged in soil survey activities in the various States at this time. During 1993 the increased testing activity (including field workshops) resulted in three sets of amendments being distributed. In 1994, testing continued via published soil profile descriptions and other data bases, and frequency distribution tables for all hierarchical levels of the classification were derived from the data base. These enabled an assessment of the relative importance of the various classes, in particular at the subgroup level.
The Testing Procedure
The creation of a soil classification scheme essentially involves the erection of a tentative framework and testing it, preferably in the field but also via profile descriptions. The basic test for any classification is that the variance within classes must be less than that between them. Perhaps the simplest test is to see if you end up with very different soils in the same pigeon hole, with only the keying properties in common. In all classification schemes it is hoped, sometimes assumed, that there is a degree of covariance between the keying properties and those you wish to predict. Unfortunately, experience has shown that the degree of covariance between some soil properties is either low or not well established. This particularly applies to prediction of various chemical and physical properties from conventional soil morphology.
The original testing procedure was one of continual modification leading hopefully to improvement, and although the diversity of Australian soils is probably finite, the law of diminishing returns also applied here. Over the period concerned Ray Isbell personally described and classified in the field in excess of 1000 profiles in all States. The use of these and soil profile descriptions in databases and in publications dating back mostly to the early 1940s enabled the creation of the original classification database of 14 000 profiles, many of which had accompanying laboratory data. The data gave a good indication of the representativeness or otherwise of the data set used to test and modify the classification.
No data was recorded for Anthroposols.
In the first edition there was an apparent bias towards Queensland, but this merely reflected the much greater availability of good quality soil profile data over many regions of this State. Considered on an area basis, the number of profiles classified per 1000 km2 ranged from 17.5 for Australian Capital Territory to 0.6 for Western Australia, with Tasmania 7.2, Victoria 5.0, Queensland 3.8, New South Wales 2.8, South Australia 1.5 and Northern Territory 0.6. However, it was not so much a matter of how many, but how representative were the profiles classified.
At the time there were large areas of Australia for which little or no soil data was available. These included, in general terms, the Northern Territory south of Daly Waters, but excluding the area around Alice Springs. Similarly, data were also scarce in approximately the northern three-quarters of South Australia. In Western Australia there are large areas with little or no available data, essentially east of a very approximate line joining Esperance and Port Hedland, but excluding the Kimberley region, the Nullarbor Plain and the southern part of the Great Victoria Desert. All of these are arid, and thus the lack of data was not surprising. However, there were also unexpected areas where data was sparse in spite of relatively intensive land use, e.g. significant parts of the Murray-Darling Basin, although soil surveys were currently in progress.
The original data also reflected to some extent the distribution of certain major Australian soils. Thus the Calcarosols are most prominent in South Australia and the Vertosols in Queensland and New South Wales. The percentages of soils with accompanying laboratory data obviously reflect the agricultural importance of some soils, but this was often confounded by different preferences between States in relation to laboratory analyses.
In spite of some deficiencies shown up by this analysis, it was thought at the time that this sample of the Australian soil population was reasonably representative of Australian soils as a whole. Certainly it was much more so than the data available for earlier classification systems. Obviously more data would have been desirable for Anthroposols and Organosols, and to a lesser extent Podosols and Rudosols. Even so, it was thought that the available knowledge of these soils (with the exception of Anthroposols) was adequate for the purposes of classification. With regard to the large areas of arid Australia indicated above where knowledge is scanty, there was sufficient indication from adjoining regions that the major soils in these areas are likely to be dominated by Tenosols, Rudosols and Kandosols of a kind common elsewhere in the arid zone.
Since its publication in 1996, the Australian Soil Classification has been widely adopted and formally endorsed as the official national system. Responsibility for maintenance and updating now resides with the National Committee on Soil and Terrain (NCST). The NCST also oversaw the Australian Collaborative Evaluation Program (ACLEP). It published a supporting volume (Concepts and Rationale of the Australian Soil Classification (Isbell et al. 1997), a CD entitled The Australian Soil Classification - An Interactive Key (Jacquier et al. 2001) a revised edition of the classification in 2002 and a second edition in 2016.
Following recommendations by the Australian Soil Classification Working Group, in 2016 the NCST released a sSecond e Edition to accommodate new knowledge and understanding of soils containing sulfidic materials. Included in these changes was the introduction of subtidal and subaqueous soils. Soils with abundant ironstone gravels were also accommodated by a Sesqui-Nodular Suborder in Tenosols.
This third edition introduces a new Soil Order, the Arenosols, which represents the deep sands that were formerly classified as Tenosols, Rudosols and Calcarosols. These latter Soil Orders have extreme morphological diversity and since the first edition have been subject to extensive investigation, particularly in South Australia, south-west Western Australia and the Northern Territory. The main changes to the third edition are described in Appendix 7.