The GIS section of the Tasmanian Department of Primary Industries, Water and Environment was contracted by the Bureau of Rural Science to create a land use dataset for Tasmania. Land use data has been identified as essential in managing the range of land degradation issues occurring in Australia. The project ran for 18 months from September 2001 to March 2003.

The initial phase of the project involved collection of ancillary data relevant to the project and the selection of suitable SPOT satellite imagery. Key ancillary data sources were obtained and converted for use in Genamap GIS, including vegetation mapping, cadastre, tenure, reserves, planning schemes, irrigation data, and a few sources of agricultural data. The project was strongly equipped to map categories 1, 2, 5 & 6 of the ALUM classification, with there being a dearth of data available to assist with mapping categories 3 and 4. Exceptions 3 to this include adequate plantation data, and excellent dataset of dairy location (useful in identifying irrigated pastures), a vineyard layer containing approximately 50% of vineyards, and several vegetation layers identifying areas without woody vegetation data (approximating modified pastures). A database and dedicated land use GIS interface was developed at this stage by DPIWE GIS staff Ross Lincoln of the University of TasmaniaÃÃÃÃÿs Central Science Laboratory undertook selection of SPOT imagery, with my assistance. This proved to be an excruciating process of perusing the existing ACRES catalogue, and waiting for some cloud free weather to enable collection of real time SPOT data. Finally a selection of 15 SPOT scenes were selected. Scenes were chosen to be as recent as possible, with a low viewing angle, high sun elevation (to minimise the effect of shadow), and most importantly low cloud/cloud shadow cover. All but one image was selected from the SPOT 4 platform, containing 20-metre resolution and 4 spectral bands. The exception was the King Island image for which the only available cloud free image was from February 1998, and the SPOT 2 platform with three spectral bands. Unsupervised classifications were carried out on the first 4 images to arrive, and were conglomerated into classes approximating vegetation, water bodies, urban areas, irrigated agriculture and dry agriculture. These were used to assist in identifying land uses, but were never directly incorporated in the final dataset. I deferred instead to the BRS preferred method of seeking regional expertise to assist in accurately identifying agricultural land uses. With some trial and error, a methodology was formed. Each map in the priority areas was created as a separate 1:25 000 map by using the corresponding Tasveg map as a template to provide a bounding rectangle and projection. A Genamap OVERLAY operation was performed on each map to create a base map containing tenure, private reserves, planning scheme/s, dams, and private timber reserves. Tenure was preferred over cadastre at this base map stage because it provided all the data necessary to map reserves, Production Forestry, Mining, and Rivers, without unnecessary lines representing property boundaries and parcels etc. The OVERLAY operation minimised the amount of linework editing, but each map generally required a significant input of time at this stage to clean up slithers and overlapping polygons between adjoining features on different datasets. The planning schemes generally provided superb data to map ALUM categories 5 - Intensive Uses. Detail varied considerably between some of the planning schemes, with some of the better one being too detailed and showing every corner store and local park. Some planning schemes were dated, and used with caution. Rural residential areas were obtained both directly from the planning schemes, and occasionally from inspection of cadastre and from field mapping. Urban areas mapped from planning schemes were generally checked against imagery to ensure boundary accuracy. Reserves were tagged according to the ALUM classification as it relates to IUCN categories. Additional reserves were obtained from the Private RFA program and the Protected Areas on Private Lands Program, and mapped as 1.1.7. Areas in planing schemes identified as protected were also mapped as 1.1.7. Recreation areas were mapped as 1.1.7 if primarily native cover, or else 5.5.3. Areas mapped as Stock Routes 1.3.2 area those areas identified as stock reserves on tenure and topographic maps. Little use was made of the Rehabilitation category 1.3.4. Reclassifying areas under Hydro Tasmania tenure was problematic. Most of these areas were in parts of the state mapped at 1:100 000 scale, leaving little time for digitising features within these polygons. Generally areas of Hydro tenure containing power stations, spillways, hydro villages and powerlines etc were mapped as 5.6.1 Utilities ÃÃÃâ electricity generation, and areas without such facilities and containing predominantly vegetation were mapped as 1.2.2 Surface Water Supply. Private Timber Reserves (PTRs) have become a major feature of Tasmanian land use, and were classified as either production forestry or plantation forestry based on the predominance of vegetation type. PTRs occurring on agricultural areas were presumed to be destined to become plantations, and mapped as such. Large PTR containing both plantation and native 4 forests were sometime manually split into the two separate forestry classes, but only in areas mapped at 1:25 000. Private Forestry Tasmania management requested it be stressed in the data that PTRs are managed under forestry codes of practice protecting significant and vulnerable habitats. They were, however, unable to provide data other than PTR boundaries. The tenure layer provided by Forestry Tasmania provided detailed polygons of MDC Protected Areas within State forests, which have been incorporated throughout Tasmania. The separation of remnant vegetation from grazed native vegetation was significantly problematic and time consuming. I borrowed my decision rules from the Billabong Creek land use mapping project, and determined remnant areas to be ÃÃÃâ Blocks of trees greater than 25ha in size No apparent disturbance of cover Crowns of the trees are large and mature Very steep broken or rocky terrain, coastal areas Absence of constructed dams as watering points for livestock No tracks leading within the block (Source: Keith Emery - Land Use Mapping in the Upper Billabong Creek Catchment, report prepared for BRS) Tasmanian coastal heaths have generally also been mapped as remnant vegetation, as have some smaller areas surrounded by intense cropping. No 1:100 000 scale vegetation mapping was available for incorporation into this project. Occasionally remnant vegetation at this scale was sourced from Tasveg 1:25 000 mapping, and occasionally was digitised manually at a coarser scale. Grazing native grasslands was extremely difficult to determine from imagery, and was originally put in the too hard basket. Much of the last few weeks of the project has been devoted to adding grazed native grasslands in the priority bioregions, as well as relabelling many areas in the Midlands from remnant vegetation to grazing native vegetation. This is the result of late field inspections identifying that very little vegetation in the midlands is actually protected from grazing. Mapping of agricultural land uses was without a doubt the most time consuming element of the project. Dairy location and an incomplete vineyard layer were the only useful datasets available at this stage. Databases of livestock tags and tattoos proved to be unsatisfactory, as did a land use valuation database. I sourced skilled DIPWE employees who were willing to attribute satellite imagery for the north west and north east of Tasmania. No body had the time or expertise to assist in the crucial areas of the northern midlands or the southeast bioregion. For the southeast this was not really a problem, as it was predominantly a region of grazing pastures and grazing native vegetation. I attributed the northern midlands myself, and was confident to be able to extrapolate what I had seen mapped in other areas of the state. All crops were mapped as irrigated unless the attributer specifically knew otherwise. Plotting a large number of 1:25 000 imagery maps was time consuming, there was always a time delay waiting for the plots to be attributed and returned, and then considerable amount of time was needed to manually digitise this data onto the maps. This method also frequently created problems with edge matching between adjacent maps. Overall the technique proved satisfactory, and enabled intensively cropped areas in the north east of Tasmania to also be mapped at 1:25 000 scale. Difficulties also arose in the north west due to cloud contamination of the imagery, and to one crucial imagery being of fairly poor spectral quality and captured in November, a suboptimal time of year. Airphoto orthomosaics were used where required, as was some ÃÃÃâblindÃÃÃÃÿ mapping of vegetation/pasture boundary relying solely on vegetation mapping. Most features of the November image were reflecting strongly in the infrared bands, presenting difficulty separating crops from pasture. Bare areas were mapped as cropping, as they were most likely recently established crops.

All lines and polygons are tagged. Adjoining mapsheets are systematically edge matched for attribute consistency and line connection See also http://nrmdatalibrary.dpipwe.tas.gov.au/geonetwork/srv/en/metadata.show?id=822