Thursday, December 14, 2017

Three Surprises about Yesterday's Weather

Richard Feynman.  Photo: Tamiko Thiel
"The only way to have real success in science is to describe the evidence very carefully without regard to the way you feel it should be. You must try to explain what's good about it and what's bad about it equally."
- Richard Feynman
The above quote is one that every forecaster should keep in mind.  A forecast is a scientific hypothesis, and improving as a forecaster requires careful evaluation of your hypotheses without cherry picking evidence or being overly kind (or harsh) on yourself.

As a scientist and as a forecaster, I get especially excited about surprises.  Things that happen that I can't explain or didn't anticipate.  In other words, they lie outside the bounds of my hypotheses.    These provide great learning opportunities and in some cases fodder for future scientific research.

Yesterday's weather provided three surprises.  The first was the haze evident in the Wasatch Mountains above the valley pollution, as discussed in depth in the previous two posts.

The second was the snow shower activity in the late afternoon and overnight.  Yes, it wasn't much, adding up to 0.13" of water equivalent and an inch of snow at Alta-Collins, but it was more than I expected.

The third was the stirring and partial mixout of pollution in the Salt Lake Valley.  Although it is faint, I can actually see Lone Peak from my office right now and observations from Hawthorne Elementary show a significant decline in PM2.5 concentrations overnight.  Great news, although air quality remains moderate.
Source: DAQ
The question then becomes why were these surprises.  In the case of the smoke from California, it was a case of inadequate situational awareness.  The smoke had clearly been blowing offshore and northward along the Pacific Coast and the thought of it actually moving inland never really crossed my mind until I was skinning yesterday morning.

The other two surprises, the snow and the pollution mixout, are related to the intensity of the short-wave trough that was dropping down the back (eastern) side of the ridge.  These are both related to the intensity of the trough, which in forecasts brushed by Utah, kept us in northerly flow at 700 mb, and dropped crest-level temperatures to only -3ºC.

Instead, the trough was further west and stronger.  It brought in more humidity, stronger flow, and dropped 700-mb temperatures to about -6ºC overnight.

This led to the somewhat more productive snowshowers, but also the cooler temperatures aloft and stronger flow helped provide some ventilation for the valley.

Could this have been anticipated?  Perhaps as a low probability possibility.  For example, I went through several SREF forecasts and could find a member or two that put out more than .10" of precipitation at Alta.  An example is below.

We don't currently mine the ensembles for guidance related to pollution mixout (my bias is snow), but perhaps if we consulted the full range of model solutions, we could find a few with a sharper trough.  The European center model, for example, had a bit of a sharper trough than the GFS and might have provided a bit better guidance.

Source: Weatherbell
Looking over all the data, chances are I wouldn't have expected the mountain precipitation and pollution mixout as a high probability outcome.  They would have been low probability outcomes, but in the range of possibilities nonetheless.  All of this illustrates the importance of assessing the full range of possibilities and not zeroing in on one forecast possibility.  After all, we live in a chaotic world.

All of that being said, I'm glad we got an angry inch of snow and a pollution stirring.

No comments:

Post a Comment