1. Introduction
The Norwegian Joint Rescue Coordination Centres (JRCC) handle more than 1500 maritime incidents each year in the Norwegian Sea and surrounding waters. Of these incidents, a substantial part involves both search and rescue1 (SAR)2. This was the motivation for developing an operational search and rescue model that could be initiated with a minimum of information and that would rapidly return search areas based on prognoses of wind and surface currents.
Maritime search and rescue is essentially about estimating a search area by quantifying a number of unknowns (the last known position, the object type and the wind, sea state, and currents affecting the object), then compute the evolution of the search area with time and rapidly deploy search and rescue units (SRU) in the search area. This puts certain constraints on the model. First, the degrees of freedom must be limited to allow easy operation. This means that uncertainty about the last known position and assumptions on the shape of the object must be tractable for operational users in real time applications. Second, environmental data (wind and current fields, either prognostic, observed or climatological) must be available in real time and third, the model must be fast enough to make it an instrument for operative search area planning.
The motion of a drifting object on the sea surface is the net result of the balance of forces acting on it from the wind, the currents and the waves. In theory it is possible to compute the trajectory of an arbitrary drifting object given sufficient information on the shape and buoyancy of the object, the wind and wave conditions, and the surface current. In practice, the net motion is difficult to compute due to the irregular geometry of real-world objects. Simplifications must be introduced to make the problem tractable, and with these simplifications errors will also be introduced.