4.3. Conclusion
We have presented a new operational model for the evolution of search areas for drifting objects. A taxonomy of common SAR objects has been set up based on the field experiments conducted to date.
A new method for decomposing and perturbing the leeway of the object in downwind and cross-wind components has been employed, yielding more robust computations at low winds and more realistic ensemble perturbations. It is found that this new method makes search areas inflate at approximately 25–50% of the rate found using older methods. The stochastic particle trajectory approach employed for this model thus leads to a significantly lower rate of expansion of search areas compared with other models.
The ensemble trajectory model is operational and can forecast search areas up to 60 h ahead in time. A seven day archive of wind and current fields allows simulations to be started as early as one week ago. This is important, as incidents are not always reported immediately.
It is found that particles are dispersed primarily through the perturbation of leeway properties, as these remain constant for each individual ensemble member. The added dispersion from perturbations (random walk) of the wind field and the current field is negligible in comparison. Going to a higher-order stochastic particle model is found to make very little difference to the rate of expansion of search areas. Thus a random walk model for the perturbations of wind and current fields is sufficiently sophisticated to capture the evolution of search areas given the high dispersion caused by the time-invariant leeway coefficients.
The vast majority of rescue operations at sea occur close to the coast. We have argued that further increasing the horizontal resolution of the ocean model will allow the model to assist searches even nearer the shore and in major bays and fjords while at the same time making the ensemble spread more realistically as eddy motion becomes more adequately resolved.
More field work is required to adequately model the various SAR objects frequently found in Norwegian waters and elsewhere. Furthermore, the experimental error of some of the older leeway categories is so high that it is desirable to revisit these SAR objects by conducting new and improved field experiments.
We have a limited, yet convincing set of field trials where the model has been compared with trajectories and end positions of real SAR objects (life rafts). The model succeeds in capturing the features of the trajectories of several drifting objects. Further studies will be required before definite conclusions can be drawn about its forecast skill, but the results so far are promising.