Sunday, November 25, 2018

A Look Back at the Model Guidance

I thought it would be worth taking a quick look at how our downscaled NAEFS ensemble forecasts did for the storm-total precipitation over Thanksgiving weekend. 

Below are forecasts plumes from the 0000 UTC 20 November downscaled NAEFS forecasts for Alta-Collins, Brighton, and the Canyons Lookout observing site.  I've added red bars to each of these showing the Utah Avalanche Center reported water equivalent and snowfall amount in upper-little Cottonwood, Big Cottonwood, and the Park City area mountains. 

The result is a bit of an apples and oranges comparison since the forecast is for a point and the observed amounts are from several sites, but there's no data available for snowfall and precip at the Canyons Lookout site and for snowfall at the Brighton site.

For the most part, the observed water equivalent and precipitation fell within the ensemble range at all three sites, although the lower water and snowfall totals in Little Cottonwood were just outside of it.  Observed water equivalent and snowfall at Brighton and Canyons Lookout were very close to being centered in the middle of the ensemble plume.  By and large, these are good results and certainly far better than what one would get using the low-resolution NAEFS without downscaling. 

That being said, one should not expect observations near the center every time.  The purpose of an ensemble is to produce reasonable probabilities for event sizes and sometimes lower probability, outlier events are going to verify.  Nevertheless, one likes to see the observations fall within the ensemble range and for the ensemble to produce a plume that is as narrow as possible. 

I've been a little concerned that the downscaling is producing too much precipitation for Alta Collins.  This is a hypothesis, but for that site, I'm thinking the downscaling might be producing too much in all but post-frontal convection events with northwesterly flow, when I suspect the downscaling will underdo it. 

One other thing to keep an eye on is the experimental GFS-FV3, which will replace the current operational GFS later this winter.  I haven't had time to do a proper look at each model for the entire period, but did look at how they did for Friday when the weather was drier than advertised and noticed that the GFS-FV3 (top) was wetter than the operational GFS in our neck of the woods. 

The FV3 is available on, so keep an eye on it and post comments when you see how it does.

No comments:

Post a Comment