Impact of Scale and Quality of Digital Terrain Models on Predictability of Seabed Terrain Types

Abstract

The local and regional distribution of seabed terrain types depends on factors such as slope and terrain ruggedness. Digital bathymetrical models (DBM) are therefore essential for mapping and predictive modeling of marine habitats. DBMs vary considerably with respect to scale and quality, and this variation is likely to influence the predictive ability of marine habitat models built with use of DBM-derived environmental predictor variables. We studied the interpretability of seabed terrain types from four different DBMs in two ways: by visual inspection of hillshade representation of the DBMs and by subjecting sets of variables derived from each of the four DBMs to maximum entropy (MaxEnt) predictive modeling of six seabed terrain types. Our results show that DBMs based on multibeam echo sounder data, resampled at resolutions up to 50 m, enable identification of concentrations of seabed terrain types characterized by moraine-ridge assemblies and plough-mark fields. Variables derived from models interpolated from contours of existing marine maps and bathymetric databases have strong limitations and merely enable separation of terrain dominated by rugged bedrock seabed from soft sediment flats. Interpolated DBMs are currently the only type of bathymetrical data that cover all Norwegian coastal waters. Their poor quality is an important obstacle to predictive modeling and classification of seabed habitats.

Keywords:

• Digital bathymetric models
• predictive distribution modeling
• spatial scale
• bathymetric resolution
• bedform features
• seabed terrain types
• Norway

Acknowledgements

This work was funded by the Research Council of Norway project MarModell and the Norwegian Defence Estates Agency (NDEA). We thank Oddvar Longva (Geological Survey of Norway), which was partner in MarModell, for assistance with data acquisition and analysis of bathymetric data in the Sandøy area. Also thanks to Stephan Blumentrath who helped in making the DBMs of the Bømlo and Herdla areas based on the original multibeam data.