| Popis: |
There is a need for fit-for-purpose maps for accurately depicting the types of seabed substrateand habitat and the properties of the seabed for the benefits of research, resourcemanagement, conservation and spatial planning. The aim of this study is to determinewhether it is possible to predict substrate composition across a large area of seabed usinglegacy grain-size data and environmental predictors. The study area includes the North Seaup to approximately 58.44°N and the United Kingdom’s parts of the English Channel andthe Celtic Seas. The analysis combines outputs from hydrodynamic models as well as opticalremote sensing data from satellite platforms and bathymetric variables, which are mainly derived from acoustic remote sensing. We build a statistical regression model to makequantitative predictions of sediment composition (fractions of mud, sand and gravel) usingthe random forest algorithm. The compositional data is analysed on the additive log-ratioscale. An independent test set indicates that approximately 66% and 71% of the variabilityof the two log-ratio variables are explained by the predictive models. A EUNIS substratemodel, derived from the predicted sediment composition, achieved an overall accuracy of83% and a kappa coefficient of 0.60.We demonstrate that it is feasible to spatially predict the seabed sediment composition across a large area of continental shelf in a repeatableand validated way. We also highlight the potential for further improvements to the method. |
