The dataset is a digital polygon coverage of Tasmania, (captured at 1:25 000 scale), detailing seven broad categories (groups) of forest vegetation, which have been aggregated from photo-interpreted forest types (PI-types) derived from varying scales of colour aerial photography. Photo-interpreted forest types (PI-types) are coded descriptions of forest vegetation. They classify forest into broad species groups as well as descriptions of stand structure. Full detail of PI-typing is described in Stone(1998). The currency of the FORESTGROUP data varies by tenure. Data for public land, (particularly State forest), is current as at June 2015 with regard to the latest land clearing, regeneration, selective logging, and planting information. The information for other native forest is current to the date of the photography from which it was interpreted; the state was covered by a 20 year remapping cycle and is now updated from Lidar and field mapping. For private land, adjustments for land clearing, regeneration, logging and planting are as advised by major forestry companies to Private Forests Tasmania (PFT) at various times to December 2021; recent changes which are not forestry-related or not associated with the activities of major companies may not be reflected in the data. First version published 2012.

1. Base Data The original dataset collected in 1995/96 was generated from colour aerial photography of the whole state of Tasmania from which new PI-Type were derived. Generally the photography was captured at a scale of 1:20 000, but areas of National Park in Western Tasmania were captured at 1:42 000. Forest type boundaries were marked onto the photos, based on stereoscopic interpretation, by Forestry Tasmania’s Resources Branch. These forest types were then coded with Forestry Tasmania’s standard classifications indicating type of forest, height, and density. See Stone, 1998. Using PHOTOGIS software, the boundaries were digitised onto the associated TASMAP 1:25 000 mapsheet base, and coded. The PI-Type codes are verified using the Java program PIDL which in turn calculates FORGROUP, based on aggregation rules. 2. Annual Updates The original PI-Types are updated annually based on information supplied by forestry companies, field information collected from regional Private Forest Tasmania offices, remapping based on Forest Practices Plan coverpage information, digitising by means of aerial/satellite imagery interpretation and modeling from LiDAR. Generally these updates focus on areas identified as having significant forest boundary or crown density changes, typically caused by human-induced events such as logging, thinning, planting and clearing, though significant wind and fire damage is also mapped where detected. Updates to reflect any height growth since 1995/96 have not been applied except where the previously mentioned changes have been detected and updated for specific locations.

Position Accuracy: The estimated positional accuracy of PI-type boundaries varies according to the method of data capture and/or scale of the data source, as indicated below: DATA CAPTURE POSITIONAL ACCURACY 1 : 20000 data capture +- 10 metres 1 : 42 000 data capture +- 21metres GPS measurement with mapping standard 12 channel receiver, eg +-5 metres ProXR Attribute Accuracy: The forest type classification is based on forest management and inventory considerations rather than floristic communities. As such PI-types are primarily structural in their description of vegetation, though they have a limited floristic view of the canopy vegetation, i.e. eucalypts are distinguished from myrtle, rainforest trees, acacias, scrub species, and groundcover groups. By its nature, PI-typed data is primarily remote-sensed i.e. only what can be determined from photographs or satellite imagery can be coded, (though ground-checking is a standard procedure for quality control). Some types are, however, derived from ground observation, particularly those representing regeneration, clearing and planting. Codes within each forest type are validated by using the PI-type Decoder Library Program (PIDL), which also calculates the appropriate FORGROUP code. Attribute accuracy is checked visually at the time of data entry. The minimum patch size is 3 hectares with all smaller areas being absorbed into neighbouring polygons. The coverage is maintained on a continual basis. It is frequently used by clients in the field who report errors. On public land over a 20 year remapping cycle, each region of the state was progressively re-photographed and new photo-interpretation was carried out. Intermittent photo-interpretation is carried out, as needed, from updated photos from smaller aerial photography and Lidar data capturing projects. On private land areas are scanned for change using recent satellite imagery. Logical Consistency: Logical consistency tests for correctness, consistency and completeness were conducted on this dataset before insertion into the GIS library, i.e. visual checking of source data; a test of item structures and definitions against the library index; a test on the completeness of valid codes and polygons; verification of table, column and item identities and definitions along with associated map tolerances; topological consistency checks to detect flaws in the spatial data; edge matching of 1:25 000 digital maps to maintain coherence in codes and lines from one map tile to another where year of photography are the same. Completeness: PI-type classification is complete for each 1:25 000 mapsheet covering the mainland of Tasmania and offshore islands. Additional: Attribute metadata in forestgroup_attributes.htm. The history, process, uses and content of Forestry Tasmania's PI-Typing are described in: Stone, M.G. (1998) Forest-type mapping by photo-interpretation: A multi- purpose base for Tasmania's forest management. Tasforests 10:15-30