National Wilderness Inventory
Australia: Our national stories
Australian Heritage Commission, 2003
ISBN 0 642 23561 9
3. The Primary Database (continued)
3.3 The Primary Data
3.3.1 Data Preparation
Where digitizing and data preparation have been undertaken directly by the NWI survey team, standards and procedures adopted have been matched to the requirements of the NWI program.
Digital primary data of cartographic precision is not required for NWI baseline purposes. The resources available to the NWI did not allow the creation of digital data sets that meet the levels of precision and accuracy required for cartographic purposes.
Developing appropriate digital data preparation procedures was a delicate balancing act given resource constraints, the large amount of digital data preparation required, and the necessity to produce a national wilderness quality database within a set timetable. Much primary data for the NWI baseline have been derived from published land information sources.
Where digital information has been prepared by the NWI, details of the digitizing process have generally been maintained on standard record sheets held at the Australian Heritage Commission. A record of the source of primary data features is also included within a source item attached to features within primary database. Record keeping of this kind was introduced progressively during the course of baseline survey work. It was not established during earlier phases of the baseline program in Victoria, Tasmania or Cape York Peninsula.
The attributing of source information for features within the primary database was introduced more recently, and has only been used thoroughly for survey work in New South Wales and parts of Western Australia. In these study areas primary data features are individually sourced. For areas surveyed prior to the introduction of source coding, codes are generally allocated on a survey unit by survey unit basis. Particular features have been sourced retrospectively where this has been possible.
Note that source coding refers specifically to the feature type and location. Because primary data features have been subject to NWI code allocation, and some form of checking and review, attribute coding is not always consistent with source material. Further details are provided in the following sections of this part of this handbook.
The digitizing process undertaken at the Geography Department of the University of Adelaide was in accordance with the guidelines and principles adopted by the Information Systems Branch of the South Australian Department of Environment and Planning (Goodwins et al 1988).
The land cover primary data set is of fundamental importance to the NWI process. It is the data set which is used to establish areas for wilderness quality survey. It is also vital for the calculation of several wilderness indicator values. Unlike line and point data, there has been no one principal data source for land cover attributes. The process used to compile land cover data varied considerably from region to region.
The various procedures used are described in detail in the appendices, while details on the land cover feature coding used in compiling this data set are outlined in Table 3.3. Four broad land cover categories are essential for wilderness analysis. These are natural cover, cultural cover, ocean, and built up areas. Data for two of these categories can be prepared fairly readily from standard topographical mapping, at the very least. However, discrimination of natural from cultural cover is problematic.
There is a very poor information base from which to work. There are not, as yet, comprehensive national data sets which can be used to discriminate these cover classes. There are also severe definitional problems in distinguishing concepts of 'natural' as opposed to 'cultural'.
Rapid change in land use and clearance patterns in the landscape simply compounds these problems. For NWI purposes, natural cover was defined as follows.
Natural Cover: an area of land or water which essentially retains its pre-European cover. Where this cover is vegetated, the structure and taxonomic composition of native vegetation communities should essentially reflect those present at the time of European settlement.
A working objective adopted for mapping purposes was to distinguish natural from cultural cover on the basis of:
evidence of deliberate efforts to clear native cover for urban, agricultural, water supply, and pasture improvement purposes.
Although no standard technique was adopted for preparing this database, five general categories of information were used:
- assembled vegetation mapping provided by project associates and collaborators;
- map preparation using large-scale air photography (completed in parts of southern SA);
- map preparation from visual interpretation of hard copy satellite imagery (completed in areas in Qld, NSW, NT and SA);
- classified digital imagery of remnant woody cover and cultivated areas (available areas in NSW); and
- map preparation from land cover information on topographical mapping (undertaken for areas in all states).
Even where the NWI team had direct control over the land cover mapping process, mapping specifications and decision rules varied quite significantly. For instance, there was a relatively high level of precision adopted in parts of southern South Australia. This contrasted with relatively coarse level mapping in central and western Queensland.
Despite this variation, some general comments can be made about the accuracy and precision of the land cover mapping process. Emphasis was placed on identifying larger and more substantial areas, rather than small remnants, as it is the more substantial areas where wilderness quality is likely to be more significant.
Generally, no special effort was made to map natural cover with a greater spatial precision than the baseline wilderness sampling resolution requires. (For baseline wilderness assessment purposes, natural areas are sampled at 500-metre and 1000-metre resolution). Relatively small, cleared areas within areas of natural cover have a significant effect on wilderness assessments and consequently special efforts were made to identify and map these.
More readily discriminated land cover features, such as built up areas and ocean were derived principally from topographical mapping or Geodata Topo-250K, where better information was not available. Because land cover data has been drawn from sources of varying quality, the level of detail available from some sources is not sufficient to distinguish all feature types specified in the NWI coding scheme.
Line and point features for the NWI primary database have been derived from a number of information sources and processed in a variety of ways. This variation arises because of considerable regional differences in the make-up and availability of information.
The most commonly used sources of information for line and point features were topographical data from Commonwealth government and state government mapping authorities, both in digital and paper form. Also significant for particular areas were specific data sets held by government agencies with responsibility for forestry, environment, infrastructure and social services.
Where information was derived from topographic mapping sources, a standard feature coding procedure has generally been adopted. Feature codes are outlined in Section 3.4 (Feature Codes and Grades). Line and point features relevant to wilderness assessment were not always extracted from data sources and entered into the primary database. Fence lines and cleared lines for exploration and seismic survey are the two cases in point. Both these data sets were generally excluded from the primary database and in almost all circumstances from analysis, even in instances where relatively sound information was available.
This decision was taken because of the patchiness and inconsistency of information for these features across study areas, and the fact that (in the case of exploration lines in particular) these features have a variable influence on wilderness quality. Inclusion would have created severe consistency problems in wilderness quality assessments, compromising the utility of results. For instance, exploration lines can have a substantial impact on wilderness quality if recently established and located in otherwise relatively remote and natural areas.
On the other hand, when unused over time, in resilient environments, the impact of these features may become negligible. In using the database, it should simply be noted that wilderness quality assessments do not include these features, and that appropriate allowances should be made.
Two additional points need to be noted in relation to these particular features. First, where seismic survey lines have become used as a part of a regularly used access network, an attempt has been made to include these features within the primary database (and subsequent analysis). Second, although fence line and seismic line information has in certain areas been included in the primary database, these features (with the above exception) have not been used in wilderness quality analysis.
The process of preparing digital point and line data from mapped sources generally included an initial hand digitizing and coding phase, followed by plotting, checking and editing. Edited data were then assembled and entered into the primary data library. Line checking and correction involved the editing of major spatial errors in digital line work (such as overshoots and undershoots) while leaving uncorrected those errors likely to produce a negligible impact on survey results. Attribute checking and editing procedures tended to vary from region to region, and are discussed in further detail in later sections of this handbook.
In more recently surveyed areas AUSLIG's Topo-250K Geodata formed the basis of linear primary data. Although this data set does not contain all linear data features required for wilderness analysis, it provides a good starting point for road, rail and track access features (and water courses where required). Other linear features, such as transmission lines, are not included in this data set and therefore required conventional digitizing. Point attributes associated with Topo-250K Geodata are generally inadequate for NWI purposes. The only point feature types used by the NWI were landing grounds and airfields. All other point features were compiled by conventional hand digitizing.
Details concerning the conversion processes for assigning NWI feature codes to the Topo-250K point and line data are available from the Australian Heritage Commission.
The model used to produce Biophysical Naturalness assessments in 'arid' regions assumes that the degree of biophysical change introduced into the landscape directly relates to the intensity of livestock grazing (refer to Section 4.3, Biophysical Naturalness).
The model also assumes, in turn, that the intensity of grazing directly relates to the distribution of permanent and semi-permanent watering points, the suitability of range type for grazing, and tenure. Three primary data sets are therefore required for running this model; the distribution of permanent and semi-permanent watering points and lines in the landscape (Section 3.3.3), livestock grazing tenures (Section 3.3.5), and the stratification of landscapes in terms of suitability for grazing.
This section of the handbook describes compilation of the latter primary data set. A three level classification of range type is required for Biophysical Naturalness analysis.
These three levels are as follows:
- 'Grazable': areas which are capable of providing satisfactory forage to domestic livestock and which in good seasonal circumstances are capable of sustaining at least moderate levels of grazing intensity.
- 'Marginal': areas which are capable of supporting grazing in good seasonal circumstances at generally low levels of grazing intensity.
- 'Non-Grazable': areas which are not capable of supporting domestic livestock grazing.
To enable this classification data sets were sought which contained areal units and attributes that could be used to distinguish these broad categories of grazing capability. A rather disparate group of information sources were used for this purpose, held in the rangetype layer in the primary database library manager. This data set was created from two alternative data sets, northcote and land_sys, which were used according to their suitability and availability for particular regions. The choice of data set and appropriateness for their use in 'arid' Biophysical Naturalness analysis are discussed in Section 4.3.1, 'Arid' Biophysical Naturalness.
The northcote layer is a national data set comprising soil-landscape units mapped for the Atlas of Australian Soils (Northcote 1960‚1968). This data set was supplied to the NWI in digital form by the National Resource Information Centre of the Commonwealth Department of Primary Industries and Energy. It consists of data collected and classified by CSIRO Division of Soils and mapped by the Division of National Mapping, Department of National Development.
This coverage consists of mapped soil-landscape units with map unit identification codes. Although the mapped soil-landscape units of the Atlas of Australian Soils primarily relate to physical soil characteristics, these units were considered suitable for use in broad range-type classifications for several reasons.
First, soil characteristics are a basic component of landscape character, and are one of the key elements controlling vegetation. There are, in many instances, fairly close associations between soil patterns and the distribution of plant and animal communities.
Second, the Atlas's mapped soil-landscape units consist of repeated associations of soils and landscape elements. These amount to generalized land-systems, useful for a range of landscape stratification purposes including generalized range classifications.
Third, the map units used in this data set have been used to support similar work. For instance the CSIRO range descriptions and classifications in the Kimberley region of Western Australia and in Queensland (Weston et al 1981).
Finally, for many parts of Australia, there are, as yet, no alternative environmental stratifications which can be used for this purpose. For these reasons, where no better information was available, the northcote coverage was used as the basis for establishing NWI range-type classes; this was the case in South Australia and Western Australia.
Descriptive categories of range suitability derived from the northcote coverage are held permanently within the rangetype library layer. The range type assignment process in South Australia was based on soil, landscape, and land use descriptions provided in Northcote et al (1975).
A more complex assignment process was developed for Western Australia where mapped soil units were matched to best fit with related range classifications provided in alternative studies, and classifications made in adjoining areas in South Australia and Northern Territory. The land_sys coverage is distinguished from the northcote coverage in that it is derived from a multiplicity of land information sources. However, the areal units comprising this coverage are also attributed directly with the range-type classifications required by the 'arid' Biophysical Naturalness analysis process.
These classifications are thus stored, like the northcote coverage, permanently within the rangetype library layer. The land_sys coverage was used for 'arid' Biophysical Naturalness analysis in Queensland, Northern Territory, and New South Wales. In these regions data sets were available that could be more directly associated with the required range-type classifications than was possible using northcote. Each data source contributing to the land_sys coverage was treated separately, according to the nature of the information it contained and the procedure required for extracting and assigning information relevant to the making of range-type classifications.
The data sources relied upon for compiling the rangetype primary data layer are shown in Figure 3.2.
The pastoral library layer comprises the third primary data set required for producing 'arid' Biophysical Naturalness assessments. This layer consists of land tenure classes that provide for livestock grazing, extending over all lands within the limits where 'arid' Biophysical Naturalness analysis was conducted. The way in which this primary data set is used in 'arid' Biophysical Naturalness assessment is described in detail in Section 4.3.1.
A three-level classification of livestock grazing tenure is required for Biophysical Naturalness analysis. The three attributes are as follows.
- 'Sheep': Areas where, within the previous (60)* years, land tenure has provided for the grazing of domestic livestock. Predominantly sheep.
- 'Cattle': Areas where, within the previous (60)* years, land tenure has provided for the grazing of domestic livestock. Predominantly cattle; cattle and sheep.
- 'None': Areas where tenure has not provided for the grazing of domestic livestock within the previous (60)* years.
* The time period, 60 years can vary from region to region.
The layer was constructed from three information sources. The main source was the Australian Public Lands Database comprising digital public tenure information compiled at a nominal scale of 1:250 000. Two supplementary sources were also used; a digital pastoral tenure database for South Australia provided by the South Australian Department of Environment and Planning, and a Digital tenure database of the Northern Territory provided by the Conservation Commission of the Northern Territory. The data sources relied on for completing the pastoral primary data layer are shown in Figure 3.3.