Thursday, 13 January 2005
Predictability as gleaned from recent Eta Model results: can we still significantly increase synoptic-scale NWP skill out to several days?
Having a nested limited area model (LAM) with lateral boundary information prescribed from forecasts of a global model, it may be expected that "the contamination at the lateral boundaries … limits the operational usefulness of the LAM beyond some forecast time range" (Laprise et al. 2000). With the Eta at NCEP run operationally out to 3.5 days for several years now, efforts are possible to look for signs of this taking place. A number of efforts in that direction were mounted by comparing the skill of the Eta vs that of its driver model as a function of time, and no signs of deterioration of the Eta relative skill were obvious. Possible explanations for this resilience of the Eta to inflow of the less accurate LBC data are looked for, and it is hypothesized that the eta coordinate is a strong candidate for a lead role in that sense. On the other hand, it has repeatedly been stated in recent years that the eta coordinate is "ill suited for high resolution prediction models"; thus, should one stay away form the eta as we are increasing resolution and trying to improve our skill? Comparisons of the Eta precipitation scores against those of a higher resolution sigma system model are shown to strongly favor the eta, and thus offer an opposite advice. The eta downslope windstorm problem, the reason for the negative eta statements referred to, is therefore revisited; it is demonstrated that the problem can be removed by a refinement of the eta discretization. Overall, it is suggested that considerations made and results shown support the expectations of Lorenz (1993) that "the coming improvements in forecasting may have to come from better representation of structures that are supposedly already resolved, or better formulations of some of the physical processes", with reach rewards likely not only from the second part of this quotation as usually assumed, but from the first one as well.