Wednesday, February 18, 2015

Predictable or Not Predictable, That Is The Question?

After a warmup for today and tomorrow, it's fairly predictable that Mother Nature will be bringing cooler weather to northern Utah this weekend, but determining how much snow she will bring to the mountains, aye, there's the rub.

We have seen this movie a few times in the past several weeks.  An upper-level trough digs into the western United States on the downstream (eastern side) of the long-wave "death ridge."  The models do a reasonable job with the long-wave pattern, but the challenge predicting the strength and structure of the upper-level trough and how it impacts precipitation over northern Utah.

There are a number of ways to examine the difficulties of forecasting an event like this. One way is to examine a series of model forecasts all valid at the same time [in the loop below, 1200 UTC (0500 MST) Saturday].  Note how in the upper panels there is considerable variability in these model solutions with regards to the structure and strength of the upper-level trough.  This in turn leads to quite a bit of variability in the coverage and intensity of precipitation, a situation that is further exacerbated by the fact that this is a pattern in which the precipitation tends to be somewhat scattered, making specific local prediction more difficult.

Another perspective is provided by examining a forecast ensemble, in this case the North American Ensemble Forecast System initialized at 0000 UTC (1700 MST) yesterday (yes, I know this is a bit old, but the 1200 UTC run won't be available for some time). The lower right hand panel is a spaghetti diagram, basically a plot of two selected 500-mb height contours from each of the NAEFS ensemble members.  If those contours overlap or are very close, the ensemble members are in agreement.  If they are spread apart, that's a sign of uncertainty.  Note how there is generally good agreement with regards to the long-wave pattern with a ridge over the eastern Pacific and the trough over the interior west, but quite a bit of spread with regards to the position and strength of the trough.
Source: NWS
If we then take the precipitation output from the NAEFS and downscale it based on local precipitation climatologies, we can get a forecast for Alta-Collins.  As shown in the plot below, there is considerable "spread" in the ensemble members.  A few produce very little precipitation ( .25 inches of water equivalent or less) whereas there are a few others that produce over an inch.  There are also differences in when that precipitation falls.  The average is a total accumulation of about .75 inches.  I'd probably shift these accumulations down just a bit since this is a situation where I expect our downscaling technique might be overdoing it a bit.

Of course, the best ensemble to examine is the ECMWF ensemble and it seems to be in the ballpark of the NAEFS.

With such a wide spread, it's quite difficult to pick a winner at this time.  In my view, the odds of 3-6 inches total accumulation through Sunday in the upper Cottonwoods is perhaps as good as rolling a seven with two dice, but as in craps, other outcomes are possible.


  1. Those ensemble qpf graphs are pretty neat. Can you access that for other grid points? If so, how?

    1. We're producing those internally for now. Hopefully will get them accessible at some point. We can produce them for anywhere in the west.