Bayesian estimation of mean shipping densities from multiple data sources

Maintaining continuous tracks of all vessels crossing a large oceanic area is a difficult task to perform. For this reason, the true mean shipping densities over a given period of time can rarely be directly calculated; they need to be estimated from a limited number of observations made at discrete points in time. The completeness and correctness of these observations may vary from one data source to another. This paper proposes a Bayesian fusion method for producing shipping density maps of high resolution based upon vessel position reports obtained from various sources of maritime traffic data. An example involving data from three sources with different characteristics is also presented.