Access to data and documentation files via a registered zip file containing 3 folders and a readme file explaining what is in each folder: 1.input data (training and test data) 2.DEM data 3.documentation about the method and validation
The point data brown_1.txt, brown_2.txt, brown_3.txt and brown_4.txt described by the metadata record 'Bathymetric surveys of Brown Bay, O'Brien Bay and Newcombe Bay in the Windmill Islands' was used as input data.
The comma delimited data resulted from bathymetric surveys of Brown Bay by Jonny Stark (AAD) in February and March 1997 includes latitude, longitude and depth fields. ... Latitude and longitude were obtained from differential GPS.
When viewing the Brown Bay soundings in conjunction with the Australian Antarctic Data Centre's coastline data and a Landsat 7 satellite image (path 104, row 107, captured 5 February 2000) it was apparent that there was a horizontal shift in the soundings. This may have arisen due to the problems with establishing the location of the GPS base station. Consequently, in consultation with Johnny Stark, a horizontal shift was applied to the soundings. The shifts applied were: easting +11.5 m, northing +21.3 m.
The shifted soundings were explored using the ArcGIS Geostatistical Analyst extension ESDA tools to identify and remove outliers in the data. Soundings that seemed irregular were deleted. Soundings that were positioned above the current coastline data were deleted although the coastline data were not used as input data for the dem.
Vertices from the Australian Antarctic Data Centre's terrestrial contour data in the Brown Bay area (Dataset_id 11, Qinfo 17) were added to the soundings. Half of this combined point data was used as a training dataset to create the DEM. The other half was retained as a test dataset to use to validate the DEM.
The creation of the DEM (0.5m cell size) was done in ArcGIS 9.2 using the Geostatistical Analyst extension, with the input data in a UTM zone 49 projection.
In Geostatistical Analyst: The input data was split into a training and validation (test) dataset. the DEM was interpolated from the training dataset using a Kriging method; the DEM was cross validated using the training dataset; and the DEM was validated using the test dataset.
Contours at 0.5m interval were generated from the DEM using the Spatial Analyst extension. Contours were smoothed and simplified using a tolerance of 1.5m . Contours greater than zero were exported to a terrestrial contour dataset called Contour_ln. Contours less than or equal to zero were exported to a bathymetric contour dataset called Bath_contour_ln.
The zip file available for download includes the training and test datasets as shapefiles, the DEM in ESRI's ArcInfo interchange and ASCII formats, files documenting the Kriging method and the validation results, and the terrestrial and bathymetric contours derived from the DEM.
Usually mapped to a Universal Transverse Mercator projection, Zone 49S, Datum WGS-84.
THE DATA ARE NOT FOR NAVIGATION PURPOSES
The DEM is stored as an ESRI grid, a dem.asc - Arc/Info GridASCII format (Grid) and dem.e00 - Arc/Info export format (Grid). The terrestrial contour and bathymetry contour data are stored in esri shapefile format.