Tuesday, December 28, 2010

The Coming Storm And Thoughts On Model Reliability

In a blog post below, from Christmas Eve, I noted that the GFS was more agreesive than the ECMWF at 168 hours and that ALL of the GFS ensemble members were forecasting precipitaion over southeast Arizona for this week. The early morning NWS forecast (made about 2 am today - Tuesday December 28th) today has precipitation POPs for metro Tucson at 70% for Wednesday, 100% for Wednesday night, and 60% for Thursday, as well as alerts for a significant winter storm above 5000 ft msl elevations. Thus, the large-scale, long range, forecasts at the end of last week appear to have been remarkably accurate. This coming event (I'm assuming of course that I can rely on the models!) illustrates two things for me: 1) how very much the large-scale numerical forecast models have been improved during the last decade or so, and 2) how model-dependent current operational forecasting has become. I personally find the increase in accuracy of the models quite astounding - something that I would not have anticipated 10 to 15 years ago. The nitty-gritty details on smaller scales remain at times elusive - as well illustrated by the extensive SUNY MAPS disucssions leading up to Sunday's great blizzard in the Northeast.
Above at top, I show the GFS ensemble mean and spaghetti chart 500 mb forecast valid at 1200 UTC this morning. The not very impressive, low-amplitude, wave in the Gulf of Alaska is the feature that is forecast to become the significant, and very cold, cut-off storm event over the western US. This feature appears, on the analysis charts, rather innocuous, yet all the models dig this wave rapidly south-southeastward and amplify it significantly. The agreement of all the ensemble members continues through 60-hours (see right panels of top and middle graphics - white line is the operational member's forecast). A further key to the evolving weather event is the significant plume of Pacific moisture (bottom panel) from west of Hawaii. The large-scale models have obviously picked up this feature, and its interaction with the digging short-wave leads to a much more significant storm event than might occur if this moisture influx were not available. All of this is quite amazing to this old-timer.
As recently as just a couple years ago, I would never have expected to see an NWS forecast for southern Arizona with 100% POPs that would be associated with an anticipated feature moving from the Gulf of Alaska. Ten to fifteen years ago my personal forecast methodology always started with the observational charts, before I would even begin to assess the numerical forecasts. No longer is this the case, since I often start looking at the model analyses in concert with, or even before, the observational charts - my how things change. I can't imagine how weather forecasting would go if all the numerical modeling centers crashed for several days. (Actually I can and will elaborate in a future post!)
Edited at 4 pm to add emphasis (red) and a couple of additional things (blue).

No comments:

Post a Comment