Evaluation of factors on the patterns of ship movement and predictability of future ship location in the Gulf of Mexico
Bay, Sophia M.
Koyak, Robert A.
Alt, Jonathan K.
MetadataShow full item record
In this thesis, we examine techniques used to predict future ship movement using historical Automatic Identification System (AIS) data in the Gulf of Mexico from April 2014. We process the data to remove outliers and identify subtracks, which are associated with trips made by a vessel between two points. A cluster analysis is then used to determine the extent to which subtrack routes segregate into groups in an area without well-defined shipping lanes. Although clustering structure does exist, it is not strong enough to support prediction modeling in line with other published work. We also examine the effects of weather and sea-state on deviations of a vessel's traveled route from the shortest (great-circle) route. Vessels vary substantially in how closely they adhere to a great-circle route. Head winds also contribute positively to these deviations. This result suggests that algorithms designed to predict the motion of vessels should take weather and sea-state into account.