Bad things happen to good models

I got pretty worked up yesterday about the best weather maps I have ever seen for Arizona, ones produced by our best numerical forecast model, the WRF-GFS, one that had so much rain here at the end of the month, based on billions and billions of calculations.  The rain in that model was due to a former hurricane (one that has not yet formed, but like me, is still merely a depressed area–in the eastern Pacific).  It subsequently develops in the model into a strong tropical storm and its remains were forecast to track right into Catalina and SE Arizona and produce as much as 1.5 to 2 inches of rain in 12 h.  I loved that forecast map so much!  Thinking about having it framed.

And, rain in various amounts, was forecast here for a couple of model runs, not just one, before going away lately.

This is what weatherfolk deal with all the time, these model vagaries in the longer forecast range of 7-15 days. In the two maps below, one from yesterday’s blog, and the one to the right of that, the model run rain map from last night’s run for the SAME time and day, the afternoon of Friday, September 28th:

Well, that “solution” is gone now (see map at right), the one having no rain at all in AZ for the same time as the flood shown in the map at left.  That remnant of a tropical storm off Baja shown on the right, just sits, spins lethargically, and dies, its moisture staying to the south of AZ now.  I guess Wildcat fans will be happy that the football game on the 29th here won’t be played in the mud with some leftover rain falling.

This rain disappearance was not surprising,  I guess, when “we” looked at the NOAA spaghetti plots and saw that the confidence level had to be low because of all the vagaries produced by “perturbing” the model with a little bad data at the outset of its run.

If I was a weather presenter, my 15th day would be stacked with the different forecasts produced by the model for that same day as it got closer in real time.  This would be the purpose of demonstrating to my TEEVEE watchers that the forecasts for the SAME day, such as September 28th, are changing drastically: “Sunny!” “Rainy, 2 inches!” “Partly cloudy!” “Rainy, a quarter of an inch!  “Sunny!”  “I don’t have a clue!”–the correct answer for display.

Don’t give up all hope for rain here, though, in the coming two weeks.

First, the unpredictablility of our current situation means that the models could calculate a rainy “solution” for us at the end of the month yet.  Second, another hurricane/topical storm comes marching up the coast of Baja at the end of the first week in October, heading in the direction of AZ.  So, all the rain potential here in our Arizona fall is not lost yet.


BTW, the uncertainty here was NOT the case for the weather back East in the spaghetti plots, where a cold regime has been solidly predicted there with considerable confidence in the 7-15 day range (forecast contours lines were pretty bunched up in that region AFTER the errors were introduced).  Low temperature records are already falling in a few locations, and more will fall over the next two weeks.



Nice pancake clouds yesterday, Stratocumulus clouds resulting from the spreading out of shallow Cumulus clouds.  Here’s an example over the Cat Mountains.  Tops ran about -10 C, maybe -11 C.  Thought I saw a very, very fine ice veil between some clouds coming off the mountains, but, it could have been an aerosol haze.  Concentrations of ice would have been VERY low, anyway, less than 1 per liter of air, in this marginal for ice-formation day.
This morning’s TUS sounding indicates a cooling at cloud top to -14 C or so.  This will mean that you should see some virga around today in somewhat fatter clouds.  Yay!  Virga.  Egad!  I just looked out the window here at first light (5:50 AM), and there it is, that virga!  There’ll be a real pretty sunrise as a result
The End.