Printversion of this page
PDF-Version of this page


Eltanin impact survey

Analyses of High Resolution Bathymetric Data in the Eltanin Impact Area


3D-View of the Eltanin impact area

DEM of high resolution bathymetric data in the Eltanin impact area.





(taken from the Master Thesis by Merijn J. Jacops):

High resolution bathymetric data are widely used for interpretation of the morphology of the sea floor, for geological, and geophysical research, and for the creation of bathymetric maps. The Alfred Wegener Institute for Polar and Marine Research maintains a multibeam system, called Hydrosweep DS-2, for the surveying of the sea floor in the polar regions of the earth.

One of these research projects, in the last years, was the Eltanin Impact Area, situated in the Bellingshausen Sea, at the western entrance of the Drake passage, in the vicinity of the Antarctic waters, and some 700 nautical miles off the Chilean coast. This Eltanin Impact Area is yet the only known impact area affecting the deep ocean basins, and the evidence for this was found in the early 1980's [Kyte et al., 1981].

During two expeditions, in the year 1995 (ANT-XII/4) and 2001 (ANT-XVIII/5a), with the German Research Vessel 'Polarstern' this area was for the first time systematically surveyed, and bathymetric data was collected within a large area. However, during these two expeditions not only the bathymetric data were collected, but also pseudo side scan data, and backscatter data. These data formed the basis for the master research, presented in this thesis. The main tasks of this research project were first the derivation of a high resolution and accurate digital elevation model of the Eltanin Impact Area, better than 1% of the water depth, within the Geographic Information System (GIS) ArcInfo, and second the backscatter processing and analyses.

For the computation of a digital elevation model different steps were undertaken. The most important step was the bathymetric data editing and cleaning, to remove the outliers, gross and systematic errors. This data editing and cleaning was performed using the software HIPS of the GIS program CARIS. The editing and cleaning procedure was split up in three parts. First the navigation editor was used to correct for navigational errors, followed by special filters within the swath editor to reject erroneous bathymetric data points. Finally a surface cleaning was performed to remove erroneous data points, based on statistical methods. These edited and cleaned data were used within ArcInfo to compute a digital elevation model (DEM). Within ArcInfo different methods for the computation of a digital elevation model were analysed. These analyses were founded on the following assumption:

The modelled surface has to fit the original data in the best possible way, with the assumption that the noise does not influence the modelled surface. This means that an acceptable smooth surface has to be modelled within the measurement accuracy.

Within ArcInfo the inverse distance weighting interpolator was used to compute this DEM with the use of a search radius depending on the number of input points, but with a maximum size of 500 meters. The grid size of the DEM was chosen at 100 meters, based on the average point density. Because of the noise that was present after the computation of the DEM, a slope depending filtering was performed to reduce these effects. For the filtering a binomial filter was used, with its size depending on the topography. In flat areas, because of the high signal-to-noise ratio, more filtering was performed than in steeper areas, where there is a lower signal-to-noise ratio. At the steepest areas no filtering was performed. This DEM was used as input for the derivation of 20 meter contour lines, and for bathymetric mapping. The mapping was performed using ArcGIS. Four map sheets were produced on a scale of 1:100 000, and two overview map sheets on a scale of 1:200 000 and 1:500 000, covering the area from 56° 50' S to 58° 00' S, and from 90° 10' W to 92° 20' W. Further a basis for a GIS of the Eltanin Impact Area was created with information about bathymetry, geomorphology, and backscatter data.

After these analyses of the bathymetric data, it could be concluded that the software used to clean these data is an important and strong tool within the bathymetric data processing. Almost 10-15% of the bathymetric data had to be rejected because of outliers, gross, and systematic errors. These rejected data points were especially found within the less accurate outer beams. After the editing and cleaning, a digital elevation model, using ArcInfo, could be computed within the accuracy of the multibeam system, being in optimal conditions 0.5% of the water depth, and within the accuracy described in the tasks of the research project, being better than 1% of the water depth. Also the assumption made before, was realised. However, it should be analysed if the accuracy of the computed digital elevation would increase when using a special weighting function for the data points depending on the positions within the swath.

During the two expeditions the multibeam system Hydrosweep DS-2 not only collected the bathymetric data, but also backscatter strengths data. These data were analysed for an interpretation of the bottom roughness and bottom type within the Eltanin Impact Area. With the use of the computed digital elevation model the two important parameters for backscatter analyses, the backscatter strengths and the incidence angle, were computed. These data were combined for response curve analyses. This is a method to differentiate between different roughness of bottoms, and bottom types, depending on the incidence angle and the backscatter strength. Within ArcInfo a segmentation was performed using an unsupervised classification. Due to problems with the backscatter data of expedition ANT-XII/4, only the backscatter data of expedition ANT-XVIII/5a were used for this segmentation.

The use of an unsupervised classification for segmentation of the backscatter data, using the response curve analyses, proved to be successful. A global interpretation of the different roughness of bottoms, and bottom types, could be made, also in comparison with the digital elevation model of the Eltanin Impact Area. In flat basins probably thicker and finer sediments are found, whereas at steep flanks and in areas with a lot of relief, coarser bottom types with thin sediment layers are to be found. However for a full view and better analyses of the backscatter data, the analyses of ground truth data, grain size and bottom type of the upper sediment layer, is very crucial.

Printversion of this page
PDF-Version of this page

Additional Information

Results of Eltanin impact survey in PANGAEA