The biotopes that are mapped in MAREANO contain unique combinations of species reflecting the particular environments where they are found. Modeling is the most cost effective way to create broad scale maps of biotopes on the seafloor. In MAREANO, we use multibeam echosounders to attain bathymetric data from which we can calculate the various terrain variables that help illustrate the varying environment variations. In addition, the backscatter from the echosounder provides information about the seafloor type that is used to compile geological seabed maps.
Geologists and biologists collaborate to produce biotope maps, following three steps. The first step is to identify and estimate the population size the animal communities from video surveys. These observations are used in multivariate analysis which serve to reveal natural groupings in the biotopes. It is a challenge to identify the most significant similarities and differences. To assist here, researchers use a second statistical analysis to find distinctions between groups based on objective statistical criteria.
Once the samples have been classified into groups, or biotopes with distinctive species composition, these classifications become "the machine-learning data" that will be used for modelling. The goal is to turn these data points into broad scale maps that reveal the likelihood of finding the distinct biotopes. Maps can only be made in areas with areal coverage of echosounder data. This data is used to identify the various terrain variables that will serve as reliable predictors of the surroundings.