17A49F2A01AAEEF9E050CD9B21447516EnglishData Requests OfficerDepartment of Lands, Planning and EnvironmentGeospatial Services Unit (on behalf of department)Datarequests.DLPE@nt.gov.au2017-07-25AS/NZS ISO 19115 Geographic Information - Metadata, ANZLIC Metadata Profile1.0 Land Types of the Southern Part of the Northern Territory2015-07-292017-03-15VERSION 3 - Published March 2017 This polygon spatial dataset provides information on land types over the southern part of the Northern Territory, giving land type boundaries and describes their dominant landform, soil and vegetation. This is a cumulative mapping project that is under development on a region by region basis. The current dataset, Version 3, containing the regions of Barkly Clay Plains, Davenport Ranges, Elliott Ranges, Tennant Creek Sandplains and South Sturt & Gulf Fall was published in March 2017. This dataset supersedes the information published in July 2015 (Version 1. Barkly Clay Plains) and April 2016 (Version 2. Regions excluding South Sturt & Gulf Fall). This dataset is mapped at a nominal scale of 1:250,000. In general, each land type is based on one or more components of one or more land systems mapped in the amalgamated dataset SOUTH_1M, (Land Systems of the Southern Part of the NT) which is in turn based on the original land systems mapped and described by CSIRO in the 1940s and 1950s. The description of landform, soil and vegetation for each land type is based on the information contained in the original CSIRO reports. In addition, where available, information also incorporated descriptions from finer scale land unit surveys if they occurred within the mapping boundaries and vegetation site information and/or road note information if it existed over the area. The polygon boundaries are derived from remotely sensed imagery with pixel resolution of approximately 30m. The boundaries were created using an automated segmentation process that groups like-pixels into polygon segments. These segments were subsequently classified into land types. CAUTION: Although field information has been collected, the final dataset has not been validated. Users are advised that attribute accuracy varies over the study area. Download the spatial data package (contains spatial data, maps and reports) via https://data.nt.gov.au/ You are licensed to use the DENR geospatial products on the terms and conditions set out in: Creative Commons Attribution 4.0 International Public License (CC BY 4.0) at: https://creativecommons.org/licenses/by/4.0/legalcode If you do not agree to the terms and conditions you must not use the geospatial products. You are free to copy, communicate and adapt the licensed material, provided that you abide by the licence terms (including Attribution) and attribute the licensed material using the statement: Supplied by the Department of Environment and Natural Resources. Copyright Northern Territory Government.English-22.36-16131.711382014-03-012017-03-15datasetnot knownComplete - Version 3. The current dataset contains the regions of Barkly Clay Plains, Davenport Ranges, Elliott Ranges, Tennant Creek Sandplains and South Sturt & Gulf Fall. This is a cumulative mapping project that is being compiled on a region by region basis. It will be updated if and when an additional region is completed.0not knownPolygons are naturally joined without overlap or gaps because they were derived from a single surface. GIS editing tools will be used to merge neighbouring regions. Polygons larger than 10ha are eliminated.0not knownMap unit boundaries were derived from an automated segmentation process of raster data at a pixel resolution of approximately 30m. Positional accuracy is thus generally considered to be high. However positional accuracy is considerably lower where map unit boundaries are diffused (i.e. between land types that appear similar on imagery and have indistinct boundaries, for example between two land types that similar dominant vegetation species). The data was created and edited using an Albers Conical Equal Area projection (central meridian 132.0; standard parallels -18.0 and -36.0). The final dataset is stored in the department's spatial library in geographics, datum GDA94.0not knownAttributes were assigned according to the dominant features of the land type, as determined from previous mapping over the area, field observations (where available) and other spatial biophysical data. Attribute accuracy is considered to be highly variable across the study area, ranging from moderate-high in land types and regions where plentiful field data occur to very low in land types and regions where there are little or no ground data or finer-scale mapping. CAUTION - Although field information has been collected, the final dataset has not been validated. Users are advised that attribute accuracy varies over the study area. Attribute values, as described in the Data Dictionary below, have been joined to polygons based on their Land Type classification.0Map unit boundaries were derived from digital satellite imagery (Landsat-8 mosaic ~ August 2013, products derived from the Landsat archive, and ALOS radar data) and digital elevation model (SRTM-derived 1 second Smoothed DEM and derived products), all with a pixel resolution of approximately 30m. Analyses were performed in eCognition and ArcGIS. Descriptions of land types were based on descriptions from original reports and refined using finer scale land unit descriptions where they overlap with the land type, and field observations where available. They were compiled according to national standards at the time of capture. Each land type is generally based on one or more components of one or more Land Systems as mapped and described in the dataset SOUTH_1M. Map unit boundaries were derived using the above digital data using an automated segmentation approach. A segmentation was performed on a composite image including Landsat-8 imagery (August 2013), vegetation products developed through the QLD Department of Science, Information Technology, Innovation and NTDLRM Collaborative Research Agreement 2015 (seasonal fractional cover, persistent green, water mask); and DEM and derived products (slope, relief). The scale factor of the segmentation was chosen such that resulting segments visually appeared to capture landscape features at a scale of 1:250,000. Segments were classified into land types using a combination of 1) an automated random forest classification based on manually selected sample segments and 2) manual classification. Selection of sample segments and manual classification were informed by field observations and finer scale land-unit mapping where available, and largely by a visual analysis of satellite imagery (including Google Earth and layers of the composite image mentioned above), elevation products (including hillshade), geology mapping and other available biophysical spatial information.