The complex geophysical environment of the western Arctic (Alaska, Siberia, Canada and the adjacent marine areas) present challenges to the current generation of weather and climate
prediction models, leading to high uncertainties in short range operational weather forecasts as well as assessments of future global change.
Recently, the use of ensemble forecasting techniques has become popular among operational weather prediction centers out to medium ranges, with an encouraging degree of skill evident in such an approach. In this paper, we investigate the utility of a "physics-based" ensemble approach for short range (2-3 days) weather prediction in the western Arctic regions. In such a "physics-based" approach to constructing ensembles, the members are derived by
varying the physical parameterization schemes used in the model simulations. Given that the sparse data coverage in the Arctic regions makes estimation of appropriate uncertainties in initial conditions somewhat problematic, our approach takes advantage of the fact that detailed initial error statistics are not needed. Instead, it is assumed that the range in physical treatments will produce solutions that span an adequate "error space" within which a suitable ensemble
mean solution can be derived.
In this paper, we examine the validity of this assumption for the Arctic and the utility of the resulting ensemble mean solutions for operational weather prediction, focusing on two case study periods, one in September 1994 and one in November 1997