Page 0 Metadata Land Use of Australia, Version 4, 2005-06 Data Set Title Land Use of Australia, Version 4, 2005-06 (final, completed June 2010) Custodian Australian Bureau of Agricultural and Resource Economics and Sciences (ABARES) Jurisdiction Australia Description Abstract The Land Use of Australia, Version 4, 2005-06, is a land use map of Australia for the year 2005-06. The non-agricultural land uses are drawn from existing digital maps covering six themes: topographic features, catchment scale land use, protected areas, world heritage areas, tenure and forest cover. Time series data at relatively high temporal resolution are available for the protected areas and forest themes. Only intensive land uses (e.g. built-up areas, mining) and plantation forestry were drawn from the catchment scale land use data. The types of agricultural land uses to be mapped and their abundances were based on the 2005-06 agricultural census data collected by the Australian Bureau of Statistics (ABS); the spatial distribution of the agricultural land uses was modelled and has largely been determined using Advanced Very High Resolution Radiometer (AVHRR) satellite imagery with training data. Irrigation status has also been mapped. Existing digital maps also contributed to the classification of grazing land as native or modified pastures. The maps are supplied as ArcInfo (Trademark) grids with geographic coordinates referred to GDA94 and 0.01 degree cell size. The distribution and abundance of most of the agricultural land uses is described by a set of floating point grids serving as probability surfaces for specific agricultural land uses. There is also a single integer grid serving as a categorical summary land use map, which has a value attribute table (VAT) with columns defining the mapped agricultural commodity groups, their irrigation status and the land use according to the Australian Land Use and Management Classification (ALUMC), Version 6 (http://www.daff.gov.au - search site for ALUM). Prospective users of the data should note that caveats and additional metadata are included in a document entitled 'User Guide and Caveats: Land Use of Australia, Version 4, 2005-06' (ABARE-BRS, 2010) and that the Version 4 map differs significantly from the Version 3 maps. Search Words AGRICULTURE AGRICULTURE Crops AGRICULTURE Horticulture AGRICULTURE Irrigation BOUNDARIES BOUNDARIES Administrative BOUNDARIES Biophysical BOUNDARIES Cultural FLORA FLORA Exotic FLORA Native FORESTS FORESTS Agriforestry FORESTS Natural FORESTS Plantation HERITAGE HERITAGE World HUMAN ENVIRONMENT LAND LAND Conservation LAND Conservation Reserve LAND Cover LAND Topography LAND Use VEGETATION VEGETATION Structural WATER WATER Lakes WATER Surface WATER Wetlands North Bounding Latitude -9.995 South Bounding Latitude -44.005 East Bounding Longitude 154.005 West Bounding Longitude 112.505 Data Currency Beginning Date 2005-04-01 Ending Date 2006-03-31 Data Set Status Progress Complete Maintenance and Update Frequency As required Access Stored Data Format DIGITAL ArcInfo 9.1 under SunOS Available Format Type DIGITAL - ArcInfo raster Access Constraint Users of this data set are asked to acknowledge, in any visual or published material, that it was derived and compiled by ABARES and to make known to ABARES any errors, omissions or suggestions for improvement. Data quality Lineage I. Six thematic layers were constructed as 0.01 degree rasters and overlain to determine the non-agricultural land uses and the distribution of agricultural land. The themes were topographic features, catchment scale land use, protected areas, world heritage areas, tenure and forest cover. The main sources were: 1:250,000 scale topographic data published by Geoscience Australia (GA) in 2006; land use data (plantation forestry and intensive land uses such as built-up areas and mining) from the collaborative catchment scale land use mapping coordinated by ABARES (BRS, 2006); protected areas and world heritage areas data compiled by the then Australian Government Department of the Environment, Water, Heritage and the Arts (DEWHA), now the Australian Government Department of Sustainability, Environment, Water, Population and Communities (DSEWPaC) c. 2007; land tenure data compiled by BRS in 1997; and native and plantation forest data compiled by BRS in 2007. II. The spatial distribution of specific agricultural land uses was modelled using area constraints based on the 2005-06 agricultural census data collected by the ABS and reported at statistical local area level. Discrimination between land uses was provided by monthly NDVI images covering the period 1 April 2005 to 31 March 2006, a total biomass image, crown cover data and slope data. NDVI images were from AVHRR data processed to correct for cloud cover by DEWHA (now DSEWPaC). The pixel by pixel sum of the twelve NDVI images served as a surrogate total biomass image. Crown cover data were from the native and plantation forest data discussed in step I. Slope data were from the Digital Elevation Model, Version 3 published by GA in 2008. Agricultural land uses (ignoring irrigation status) were mapped in non-timbered agricultural land using an algorithm called SPREAD II, developed by Simon Barry, formerly of BRS (Smart et al., 2006). SPREAD II, like the SPREAD algorithm of Walker and Mallawaarachchi (1998), uses time series NDVI data with training data to spatially disaggregate agricultural census data. The SPREAD II methodology is statistically based, using a Bayesian technique - a Markov Chain Monte Carlo (MCMC) algorithm. Training data were collected for the National Land and Water Resources Audit and relate to the four years, 1996-97 to 1999-00 (Stewart et al., 2001). To increase its discriminating power, SPREAD II was run using two spatial constraints: one identifying horticulture pixels and the other identifying cultivated (including horticulture) pixels. The SPREAD II outputs comprised probability surfaces and an agricultural land use summary map. Additional grazing land allocations were then made, outside SPREAD II, up to the area reported in the agricultural census. Allocations were to agricultural land pixels with crown cover up to 80% prioritised, first, by crown cover (lowest crown cover classes given highest priority) and, second, by slope (smallest slope values given highest priority). Finally, irrigation status was mapped, outside SPREAD II. Irrigated area data available at statistical division level from the 2005-06 agricultural census served as area constraints. A spatial constraint was also used; it identifies pixels in known irrigation areas. The area constraints were partitioned between the known irrigation area pixels and other pixels. 'Irrigated' status was then allocated to pixels to which agricultural land uses (other than agroforestry) had already been allocated by SPREAD II, prioritised by total biomass (largest total biomass values given highest priority). III. The non-agricultural and agricultural land use data from steps I and II respectively were combined in a single summary map with land use attribute values interpreted in terms of the Australian Land Use and Management Classification, Version 6 (http://www.daff.gov.au). Positional Accuracy The data type and stated positional accuracy of the major existing data sets used to determine the non-agricultural land uses and the distribution of agricultural land (as discussed in the lineage section) are as follows: . Topographic data - vector data, error less than 160m for at least 90% of well-defined points . Catchment scale land use data - mapping scales are 1:250,000 or larger implying that positional errors do not exceed 125 m. . Protected areas data - vector data, spatial errors variable from < 1m to 500m . World heritage areas data - spatial accuracy information incomplete but the contribution of this data set to the final product is negligible . Land tenure data - 250m raster data, spatial errors, in the main, do not exceed 125m . Native and plantation forest data - 100 m raster data, with many different sources; spatial errors, in the main, do not exceed 125m The input NDVI imagery and the output probability and summary grids have 0.01 degree pixel size. Therefore, spatial errors, in the main, should not exceed 1 km or thereabouts and a reasonable nominal scale for the data set is 1:2 million. Attribute Accuracy Non-agricultural land uses were assigned, initially, on the basis of existing data sets showing topographic features, catchment scale land use data, protected areas, world heritage areas, tenure and native and plantation forest cover. Specific agricultural land uses were modelled. Accuracy of assignments based on existing data sets depends mainly on the attribute accuracy of those existing data sets. Accuracy of all assignments also depends on the validity of the rules used for land use assignment. Concise statements of attribute accuracy are available for two of the existing data sets. For the topographic features data set (TOPO-250K), the range of allowable attribute errors is from 0.5% to 5% at a 99% confidence level. For the world heritage areas data set (Australia, World Heritage Areas), all available attributes are stated to have been checked and to be correct. The attribute accuracy of the other existing data sets is expected to be high, with consequent high accuracy in non-agricultural land use assignments. The accuracy of the specific agricultural land use allocations based on automated interpretation of NDVI images is variable. The probability grids give an indication of the accuracy of the agricultural land use allocations other than those mapped outside SPREAD II (some grazing land pixels and the irrigation status attribute). The rules used for land use assignment have been checked. The validity of most appears to be beyond question. Those which might be viewed as contentious are documented at the end of the data dictionary section of the user guide (ABARE-BRS, 2010). Logical Consistency The attribute combination corresponding to each land use assignment in the summary grid was tested by inspection to verify that these automated assignments were as intended and were logically consistent. Completeness Coverage and classification are complete. Verification of spatial and attribute data is complete but the data set has not been extensively tested. Verification took the form of a review, by state and territory land use mapping experts, of hard copy plots at 1:5 million scale of a near final version of the data set. Two plots were reviewed by each reviewer, one showing agricultural land uses at a high level of classificatory detail and one showing all land uses at a moderate level of classificatory detail. Criticisms were taken into account in the construction of the final version of the data set and in the writing of the accompanying documentation. Contact Organisation Australian Bureau of Agricultural and Resource Economics and Sciences Contact Position Data Manager Mail Address GPO Box 1563 Locality CANBERRA State ACT Country Australia Postcode 2601 Telephone +61 2 6272 2010 Facsimile +61 2 6272 2001 Electronic Mail Address info.abares@daff.gov.au Metadata Date 2013-01-11 Additional Metadata BRS (Bureau of Rural Sciences), 2006, 'Guidelines for Land Use Mapping in Australia: Principles, Procedures and Definitions', Edition 3, Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. ABARE-BRS (Australian Bureau of Agricultural and Resource Economics and Sciences-Bureau of Rural Sciences), 2010, 'User Guide and Caveats: Land Use of Australia Version 4, 2005-06', Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Smart, Robert, Simon Knapp, Julie Glover, Lucy Randall and Simon Barry, 2006, 'Regional Scale Land Use Mapping of Australia: 1992/93, 1993/94, 1996/97, 1998/99, 2000/01 and 2001/02 maps, Version 3', Australian Bureau of Agricultural and Resource Economics and Sciences, Canberra. Stewart, J.B., Smart, R.V., Barry, S.C. & Veitch S.M. 2001, '1996/97 Land Use of Australia: Final Report for Project BRR5', National Land and Water Resources Audit, Canberra, available online http://adl.brs.gov.au/. Walker, P.A. & Mallawaarachchi, T. 1998, 'Disaggregating Agricultural Statistics Using NOAA-AVHRR NDVI', Remote Sens. Environ., vol. 63, pp. 112-125.