The sediment on the seabed is an important factor that influences the ecology of the sea floor. If ecologists know what sediments occur at a site: whether they are sandy or muddy, or contain a lot of gravel, then they can begin to predict what organisms might occur there, and make sound assessments of, for example, the likely impact on fisheries of a windfarm development.
The BGS holds many data on seabed sediments, but if we use these to predict the sediment type at an unsampled site there is considerable uncertainty because sediments vary in space.
We have developed a method based on a statistical predictor called compositional cokriging which allows us to compute the probability at any site of finding a particular sediment type. Rather than telling the ecologist that the sediment at a site is in class ‘sand and muddy sand’, we provide a probability that it is ‘sand and muddy sand’ (and also probabilities that it belongs to other classes). That helps the ecologists to see how certain they can be about their predictions, and to identify plausible alternative scenarios.
The map on the right shows the probability of finding sediment texture class ‘sand and muddy sand’ on the seabed at sites on the UK continental shelf. Read a full account of this work. It was undertaken as part of BGS’s contribution to NERC’s MAREMAP programme.
Other examples in which we have used optimal statistical predictors to map spatial variables with associated measures of uncertainty:
Contact Dr Murray Lark for more information