Thursday, 15 January 2004: 4:30 PM
The risks and rewards of high resolution and ensemble numerical weather prediction
A large gap in skill between forecasts of the atmospheric circulation (relatively high skill) and quantitative precipitation (low skill) has emerged over the past three decades concomitant with advances in numerical weather prediction and data assimilation. The origins of this gap can be traced to the relatively large scale of features in the atmospheric circulation, which are more easily sampled by the observational network and forecast by models, compared to the small-scale details in the factors that produce precipitation. One common approach towards closing this gap has been to decrease the horizontal grid spacing of the numerical weather prediction models to try to simulate precipitation features more realistically. During the last few years, quasioperational experiments in high-resolution (defined here as 1–6-km horizontal grid spacing) numerical weather prediction have begun in many regions of the United States. Also at this time, research has begun to address the benefits and limitations of short-range ensemble forecast methods. Based on this experience, three major questions are discussed in this paper. First, is the information content from high-resolution NWP distinct from and/or complementary to that obtainable from ensembles? Second, what are the challenges that face modelers in transferring high resolution and ensemble approaches to operational fore-casting? Third, what is the optimal operational strategy, given continuing constraints on observational data availability, overall data management and computing resources?
These questions are addressed first through consideration of the extant meteorological and psychological research on the forecast process. Two examples (the Great Plains tornado outbreak of 3 May 1999 and Tropical Storm Floyd over southern New England on 16-17 September 1999) are presented to highlight these issues for disparate geographic regions and meteorological phenomena. Finally, the science and policy issues that must be addressed in order to maximize forecast potential are discussed.