Bayesian Geostatistics Modeling of Maritime Surveillance Data

Portugal has under its jurisdiction a vast area of sea territory and a geostrategic position, in which some of the most important commercial maritime routes occur. Therefore, Portugal needs to guarantee adequate monitoring and supervision of its waters, given the intense maritime activity as the exploitation of sea resources and fishing. Illegal actions concerning fishing remains a major threat to global maritime resources and are subject to regular surveillance actions from the Portuguese Navy. Based on the georeferenced data, collected under these actions, the main objective of this study is to build risk maps of infractions related to fishing off the southern Portuguese coast. With this aim, geostatistical data modeling techniques for binary data (presence/absence of infraction) are used, based on hierarchical Bayesian models incorporating a spatial latent component and time. For the estimation and prediction we use the Integrated Nested Laplace Approximation (INLA) approach combined with the Stochastic Partial Differential Equation (SPDE). This analysis may contribute for the definition of future routes for enforcement actions by the responsible authorities.