Wind Uncertainty in Long Range Trajectory Forecasts

Abstract

ABSTRACT The essential equations of oil spill trajectory modeling have not changed in four decades. The vast majority of existing trajectory models divide the slick up into pieces, called Lagrangian Elements (LE's). The models move the LE's by summing movement vectors related to three somewhat interrelated components; surface water currents, turbulent diffusion, and surface wind stress. Improvement in modeling has consisted, chiefly, of better refining these three components. While such trajectory models generate predictions, the real world is filled with uncertainty. As the spill response community seeks longer-range forecasts, spill models must be able to not only provide a best-guess prediction but also to estimate the likelihood and extent of any errors in the prediction. Wind forecast error often is the major contributor to trajectory error for longer-range forecasts because, more so than currents, wind is subject to rapid change in magnitude and direction. This paper provides a formula to identify a possible practical limit of predictability for dynamical wind models. It also describes a method to estimate probability bounds on slick location based on a selected set of past wind records that vary within a prescribed variance from the present and forecasted winds. Based on the degree of confidence in the wind forecast and the sensitivity to risk from a bad guess, the user can adjust the size of this variance. By extrapolating the set of selected wind records, longer-range trajectory bounds can be given that extend past any deterministic wind forecast and give long-range trajectory analysis to the spill responder.