Use of Remotely-Derived Bathymetry for Modelling Biomass in Marine Environments

Abstract

The paper presents results on the influence of geometric attributes of satellite-derived raster bathymetric data, namely the General Bathymetric Charts of the Oceans, on spatial statistical modelling of marine biomass. In the initial experiment, both the resolution and projection of the raster dataset are taken into account. It was found that, independently of the equal-area projection chosen for the analysis, the calculated areas are very similar, and the differences between them are insignificant. Likewise, any variation in the raster resolution did not change the computed area. Although the differences were shown to be insignificant, for the subsequent analysis we selected the cylindrical equal area projection, as it implies rectangular spatial extent, along with the automatically derived resolution. Then, in the second experiment, we focused on demersal fish biomass data acquired from trawl samples taken from the western parts of ICES Sub-area VII, near the sea floor. The aforementioned investigation into processing bathymetric data allowed us to build various statistical models that account for a relationship between biomass, sea floor topography and geographic location. We fitted a set of generalised additive models and generalised additive mixed models to combinations of trawl data of the roundnose grenadier (Coryphaenoides rupestris) and bathymetry. Using standard statistical techniques—such as analysis of variance, Akaike information criterion, root mean squared error, mean absolute error and cross-validation—we compared the performance of the models and found that depth and latitude may serve as statistically significant explanatory variables for biomass of roundnose grenadier in the study area. However, the results should be interpreted with caution as sampling locations may have an impact on the biomass–depth relationship.