How robust are your prediction derivatives? Error propagation modelling for seafloor terrain analysis of multibeam bathymetric data using dynamic simulation tools

There are many potential applications of seabed habitat mapping for which estimates of uncertainty in
seafloor terrain derivatives will provide additional crucial information. This information is useful to
assess the robustness of multibeam bathymetric data for seabed monitoring, change detection and
habitat prediction analysis.

As multibeam data collections build up over time a method to assess the compatibility to spatially
merge these data are essential. These datasets then become the input data for spatial analysis
procedures to characterise the seabed. The accuracy of seabed terrain first and second order
derivatives and their associated levels of uncertainty are extremely hard to convey visually or to
quantify with existing methodologies.

In this study Monte-Carlo simulation techniques were used to represent DEM (digital elevation model)
uncertainty and its effect on three topographic parameters (slope, curvature and aspect). Different
methods for representing error and quantifying uncertainty are investigated. In these results the
analysis the multibeam bathymetric error are assumed to be spatially auto correlated across a
neighbourhood zone, methods for the assessment of autocorrelation will be discussed. Each terrain
derivative layer was perturbed using its error model with increasing levels of error, and the effect on
the seabed map was assessed.

Quantifying uncertainty in the input data for habitat suitability modelling is imperative to establishing
methods for prediction, and monitoring, as we need to be able to separate potential mapping error
from change and variation in the system that we are monitoring. By combining bathymetric processing
and uncertainty modelling techniques we can make an important step towards identifying tools for
seabed monitoring and risk assessment for policy-making. These tools will improve our ability to
assess and communicate the accuracy of the seabed maps through spatially mapping the degrees of
uncertainty in our predictions and therefore make more informed choices of the data we use to inform
ocean management policies and subsequent seafloor analysis.

Document type: