Australian Antarctic Territory Coastline 2003

Brief description

The 'Australian Antarctic Territory coastline 2003' dataset is a digital vector representation of the coastline of Antarctica, between 45 to 160 degrees east, based on both the edge of permanent ice and grounding line, derived by means of remote sensing interpretation.

A 'proof of concept' methodology over a test area was carried out to compare a number of complementary remote sensing techniques, including interferometry and airborne ice radar profiling, to confirm validation of grounding line as mapped from Landsat 7 ETM+ imagery. This methodology concept then served to validate grounding line locations elsewhere along the coast of the AAT.

The National Mapping Division of Geoscience Australia and the Australian Antarctic Division developed this dataset as a joint project.

Where available, Australian Antarctic Division supplied large-scale vector data of various areas around the AAT, which were included as part of the main coastline dataset. These included:

* Holme Bay 1:25,000 GIS dataset
* Larsemann Hills - Mapping from aerial photography captured February 1998
* Rauer Group 1:50000 Topographic GIS Dataset
* Vestfold Hills Topographic GIS Dataset
* Windmill Islands 1:50000 Topographic GIS Dataset
* Cape Denison and McKellar Islands GIS dataset from Ikonos satellite imagery

Refer to the metadata record for each of these datasets for further information.

The coastline dataset is comprised of three parts: one polygon coverage consisting of ice features, and another one consisting of coastal features. A third coverage consists of only island point features (islands too small to be shown as polygons).

This dataset supersedes the Australian Antarctic Territory Coastline 2001 dataset which is also part of SCAR's Antarctic Digital Database (ADD) version 4 and version 5. It replaces data digitised from Landsat 4 and 5, with that from Landsat 7 ETM+, because of its more reliable positional accuracy and more recent acquisition.

The Australian Antarctic Territory Coastline 2001 dataset and metadata record have been archived. Please contact the Australian Antarctic Data Centre if you would like a copy of this data and metadata.

Lineage

Progress Code: completed

Statement: Some information was sourced from TERRA.

POSITIONAL ACCURACY:
The Antarctic coastline has been derived mostly from Landsat 7 ETM+ imagery with a spatial resolution of 30x30m pixels. Imagery dates used ranged from 1999 to 2002, making it a consistent 'one epoch' dataset. Landsat 7 ETM+ was the preferred sensor due to its improved horizontal accuracy, as well as being the most current. Subsets of some scenes were pan-sharpened to 15x15m pixel resolution over areas requiring better definition, as in the case of areas containing numerous small islands.

Images were supplied in Universe Transverse Mercator (UTM) projection and with systematic correction determined by satellite ephemeris, a satellite model based on information about spacecraft position and attitude. This information is transmitted with the imagery, or provided by the satellite operator at the time of image acquisition. It allows the imagery to be located to a limited absolute accuracy (ie. its position on the Earth).

Landsat 7 ETM+ imagery was compared to various geodetic control points whose accuracy is estimated at being less than 1m. A visual estimation of their positional accuracy against Landsat 7 ETM+ showed an average accuracy of approximately 80m or less. Radarsat SAR imagery was used as complementary data to help distinguish grounding lines. Its positional accuracy ranged between 200 and 500m. Only where any other Landsat imagery did not provide the coverage required, was it used for digitising features from. No correction was applied to these images. ASTER imagery was used in one instance, where other imagery was unavailable. Its positional accuracy is estimated at around less than 50m, when compared to Landsat 7 ETM+.

Each feature in the dataset contains a quality information (Q_INFO) attribute value relating to each of the source data. Features with Q_INFO values beginning with 'GA' were mapped from satellite imagery whilst those beginning with 'AAD' were extracted from existing AAD datasets. For Australian Antarctic Division Q_INFO values, a search can be done for data quality information (refer to the link below). Please email gis@aad.gov.au for further information about AAD datasets. Q_INFOs from Geoscience Australia can be accessed through the data quality table available for downloading (refer to the link below).

Horizontal datum of the coastline dataset is ITRF2000 with ellipsoid GRS80.

ATTRIBUTE ACCURACY:
Line and point feature attributes were assigned to each feature after digitising. The line attributes of this dataset describes the coastline types as best interpreted from the imagery.

Feature code accuracy for most well defined features, such as rock coasts, is expected to be between 95-100% accurate. Whereas, for features that are more interpretive, such as grounding lines and some glacier tongues, an accuracy of approximately 50-80% is expected. Additionally, three quality attributes rating accuracy were also added to each line feature, to reflect the reliability of the feature code attribute and feature accuracy. A quality information identifier relates to a lookup table of the data source used for capturing each feature.

Point feature codes have only one feature code type, as they denote only one feature type - islands mapped as point features, and are included in a separate coverage.

COMPLETENESS:
Spatial data coverage of both the edge of permanent ice and grounding line is continuous and complete for the entire dataset.

All line features are assigned a feature code, accuracy classification, compiler, source, revision date, data quality pointer, and a unique feature identifier. Other attributes are not necessarily filled, as they are non-mandatory.

Centroids and point features are assigned a feature code, compiler, revision date, data quality pointer, and unique feature identifier. Other attributes are not necessarily completed, as they are non-mandatory.

Arc/Info software was used to manually verify that each feature had its mandatory attribute item filled.

LINEAGE:
A decision was made to update coastal mapping along the Australian Antarctic Territory. As per existing SCAR data, both the edge of permanent ice and grounding line were mapped.

Staff involved with the project required some initial consultation with glaciologists from the Australian Antarctic Division in Hobart, learning about ice behaviour, structure and patterns to help interpret and determine the grounding line from satellite imagery. Landsat imagery was spectrally stretched and modified to various band combinations to help highlight different features that would assist in grounding line determination. Radarsat imagery assisted in the interpretation of grounding line. Satellite imagery were displayed in UTM projection using ERDAS IMAGINE 8.4 software. An Arc/Info vector coverage in geographicals (Lats. and Longs.) was then overlayed onto the imagery, and an interpretation of the coastline digitised from these. Where lines closed off into polygons (eg. islands), a centroid was placed within it. Attribute information was completed for each feature after digitising.

TOPOLOGICAL CONSISTENCY:
Arc/Info software was used to do a manual topological consistency check to detect errors in the spatial data structure and to correct them. This check ensures that all classified polygons are closed, nodes are formed at the intersection of lines and that only one label exists within each polygon.