Accuracy assessment of publicly available digital elevation models of the ocean floor, at the polygons of multibeam data coverage in the Norwegian and Barents Seas

Abstract

The accuracy of publicly available bathymetry grids such as GEBCO_08, SRTM30_Plus, and Predicted Topography is evaluated at six studied polygons of GIN RAS multibeam (MB) coverage. High-resolution and accuracy, independent GIN RAS MB grids are used as the ground truth. Comparison of GIN RAS MB grids with USCGC “Healy” and RV “Oden” gridded MB data revealed no systematic errors in the GIN RAS data. In order to assess the accuracy an algorithm was developed to calculate depth difference between grids built in different map projections with grid cells that do not match. The assessment revealed that GEBCO_08 is more accurate than S&S over three of the six polygons, which include mid oceanic ridge, shelf and continental slope areas (polygons 1, 5 and 6). The observed differences in the accuracy between GEBCO_08 and S&S are due to the source data accuracy at polygon 1, better source data coverage in GEBCO_08 for polygons 5 and 6 (MB data), and poor performance of the interpolation in the shelf areas for S&S at polygons 4, 5 and 6. These results show that morphology of the seafloor, source data accuracy and coverage as well, as interpolation method affect the accuracy of the final output bathymetry surface.