FSL Model Verification: A Collection of Case Studies for Mesoscale Circulations

FSL Model Verification: A Collection of Case Studies for Mesoscale Circulations

Forecast Research Division


Overview

On this page is a collection of fairly detailed case studies presented to document how well various models (FSL's hot-start MM5, the 20 km RUC, and the 22 km "Mesoeta") perform in predicting two mesoscale circulations common in northeastern Colorado: the Denver Cyclone and the Longmont Anticyclone. A portion of this work will be presented at the AMS Mesoscale Conference in late July 2001, and here is a pdf file version of that paper.

Index of cases

I. Longmont Anticyclone events: These first 3 cases were saved mainly for snow potential.

Very nice LGM Anticyclone case. Associated with it was a persistent area of light to perhaps moderate low-density snow mainly over the srn/wrn DEN suburbs in the aftn/evening of the 6th.

Saved the 1800 UTC runs of the hot-start MM5 and the Mesoeta.

 

B. 15 November 2000: Potential plains snowfall associated with a Longmont Anticyclone as well as mountain orographic snowfall.

Hot start MM5 runs saved from 1200 UTC and 1800 UTC on 15 November.

Shortwave trough passage with synoptic NWly flow behind it, with potential for a LGM Anticyclone and of course mountain orographic snowfall. In this case the NWlys tended to overwhelm any significant LGM Anticyclone, which was overpredicted by the model runs. A nice snowband did occur on the plains but appeared to be at the leading edge of the trough and surge of NWly low level flow, and proceeded to move from near the foothills at 13-14 UTC to vicinity of AKO by ~1800 UTC. This band was not predicted very well by the 1200 UTC MM5 run, while with the hot start the 1800 UTC run nicely captured it and held it together for awhile as it continued east (while diminishing in real life), about as forecast.

In the mountains good snowfall occurred commencing probably ~1000 UTC on 15 Nov, with most of it probably over by the evening of the 15 Nov. Both runs tended to underforecast the amount of snowfall, but appeared to do a decent job with distribution, with Steamboat (12" for 24 h ending 16 Nov at 1200 UTC) and Vail (10" for similar) getting by far the most.

 

C. 17 November 2000: Quick snow event with a northerly surge (Longmont Anticyclone) and shortwave trough in the northerly flow; also mountain orographic snowfall.

Setup was a shortwave moving south in the upper level ~nly flow with an accompanying LGM Anticyclone type surge. A similar type of feature on Wednesday produced a burst of snow more on the eastern plains but pretty much a null event many areas along the Front Range. MM5 run did not predict much with this event except for an area of snow more over zone 35 into zone 36, but very little out on the plains or even on the plains near the foothills. This despite a good job of predicting the strong turning (to upslope) winds with the LGM Anticyclone/post-surge flow. Mesoeta forecast also had precipitation predicted, but not nearly focused enough, or enough predicted. At 20 km resolution perhaps the strong e-w gradient away from the foothills would be too much to expect.

What happened was a burst of snow began ~23z locally and quickly became at least moderate snow here, with a good burst of nely winds producing some blowing snow. By ~715 pm had accumulated ~3.5" here in the parking lot, but when I got home at ~730pm could only measure 1.5", so quite an extreme dropoff with distance east of the foothills. Heaviest accumulations were in fact in the foothills themselves of zone 36, with up to 8" near Tiny Town. Note how Allenspark had less than an inch. One thing about the snow was it had very low liquid water content; at least 20:1 or higher, so in effect some model predictions were not far off from that perspective.

Saved the 18z run of both the hot-start MM5 and the Mesoeta. Most of the snow fell here in BOU from 23z/17th to 04z/18th.

In terms of the mountains, both models predicted a rather light event, and generally only an inch or two fell, with Eldora getting 3" (upslope from the east there probably responsible for most of this).

-------------------------------------------------------------------------------------------------

II. Other Longmont Anticyclone events.

      A. 30 April 2001

...description...  

    MM5hot forecasts from 1800 UTC run on 30 April 2001.
      0 h surface forecast of wind valid 1800 UTC on 30 April.
      1 h surface forecast of wind valid 1900 UTC on 30 April.
      2 h surface forecast of wind valid 2000 UTC on 30 April.
      3 h surface forecast of wind valid 2100 UTC on 30 April.
      4 h surface forecast of wind valid 2200 UTC on 30 April.
      5 h surface forecast of wind valid 2300 UTC on 30 April.
      6 h surface forecast of wind valid 0000 UTC on 1 May.
 

 

 

--------------------------------------------------------------------------

III. Denver Cyclone events.
    A. 24 March 2001

Great setup for a Denver Cyclone with a cold arctic high pressure sliding off to the east of CO creating widespread southeasterly flow surface to perhaps about 750 mb. Much warmer air above this with temps quite warm in the mountains. Concentrated on 18-24h forecasts from MM5/mesoeta/Ruc20 from 18z run on 23 March. All 3 models predicted a Denver Cyclone would develop, though there were some differences in its location and shape. Timing of DCVZ development seemed pretty close, with all the models showing a full-fledged circulation by 12-15z/24 March. Cyclone development was also predicted north of the Cheyenne Ridge and Raton Mesa Ridge.

Verification: Saved the METARS and LAPS wind analyses. A good Denver Cyclone did develop, but need to examine the details of the obs further to see which model verified best.  

 

 

 

    B. 17-18 April 2001

...description...  

 

 

    20 km RUC forecast from 2100 UTC run on 16 April 2001.
 

 

 

 

 

    C. 5 June 2001

...description...  

    Mesoeta forecasts from 0600 UTC run on 5 June 2001.
 

 

 

 

 

 

 

 

    D. 7 June 2001

...description...  

 

 

 

    E. 8 June 2001

...description... Somewhat limited case, mainly 1800 UTC stuff saved.  

 

 

 

    F. 26 June 2001

...description... Stronger sely flow, may have overwhelmed any cyclone attempts. Good time resolution of saved forecasts for first 6h, with obs and NOWRAD overlaid to verify for first 5h. NO RUC forecasts, only analyses from 1800 UTC.  

 

 

 

    G. 5 July 2001

...description... Fairly complete case with two sets of runs and verification, EXCEPT no RUC for the 18z time frame, and no RUC for the first 6 h of the 12z run (though some selected other forecasts). One nice thing about this case is that there was a nicely wrapped up Denver Cyclone, and some radar reflectivity and velocity images, with detailed surface data, were saved.  

 


Back to the FSL Home Page