Detecting suspicious activities at sea based on anomalies in Automatic Identification Systems transmissions.

Automatic Identification Systems (AIS) are a standard feature of ocean-going vessels, designed to allow vessels to notify each other of their position and route, to reduce collisions. Increasingly, the system is being used to monitor vessels remotely, particularly with the advent of satellite receivers. One fundamental problem with AIS transmission is the issue of gaps in transmissions. Gaps occur for three basic reasons: 1) saturation of the system in locations with high vessel density; 2) poor quality transmissions due to equipment on the vessel or receiver; and 3) intentional disabling of AIS transmitters. Resolving which of these mechanisms is responsible for generating gaps in transmissions from a given vessel is a critical task in using AIS to remotely monitor vessels. Moreover, separating saturation and equipment issues from intentional disabling is a key issue, as intentional disabling is a useful risk factor in predicting illicit behaviors such as illegal fishing. We describe a spatial statistical model developed to identify gaps in AIS transmission, which allows calculation of the probability that a given gap is due to intentional disabling. The model we developed successfully identifies high risk gaps in the test case example in the Arafura Sea. Simulations support that the model is sensitive to frequent gaps as short as one hour. Results in this case study area indicate expected high risk vessels were ranked highly for risk of intentional disabling of AIS transmitters. We discuss our findings in the context of improving enforcement opportunities to reduce illicit activities at sea.

Loitering with intent-Catching the outlier vessels at sea.

Illegal, Unreported and Unregulated (IUU) fishing activities pose one of the most significant threats to sustainable fisheries worldwide. Identifying illegal behaviour, specifically fishing and at-sea transhipment, continues to be a fundamental hurdle in combating IUU fishing. Here, we explore the use of spatial statistical methods to identify vessels behaving anomalously, in particular with regard to loitering, using the Indonesian Exclusive Economic Zone (EEZ) and surrounding waters as a case-study. Using Automatic Identification System (AIS) for vessel tracking, we applied Generalized Additive Models to capture both the temporal and spatial nature of loitering behaviour. We identified three statistically anomalous loitering behaviours (based on time, speed and distance) and applied the models to 2700 vessels in the region. We were able to rank vessels for individual and joint probability of atypical behaviour, providing a hierarchical list of vessels engaging in anomalous behaviour. While identification of irregular behaviour does not mean vessels are definitely engaging in illegal activities, this statistical modelling approach can be used to prioritise the allocation of enforcement resources and assist authorities under the United Nations Food and Agricultural Organization Port State Measures Agreement for management and enforcement of IUU fishing associated activities.