The Integration of WRF Model Forecasts for Mesoscale Convective Systems Interacting with the Mountai

1
The Integration of WRF Model Forecasts for
Mesoscale Convective Systems Interacting with the
Mountains of Western North Carolina Progress
UpdateJacob Carley
2
What is an MCS?

• An MCS (mesoscale convective system) is a large
  system of thunderstorms that generally persist up
  to several hours.
• These systems can span several states and are
  capable of producing large hail, damaging winds,
  and tornadoes.

3
What is the WRF model?

• Weather Research and Forecasting model (WRF).
• WRF is the new model to be used heavily in
  forecasting.
• A mesoscale model that places emphasis on
  terrain.
• Is scheduled to replace the Eta/NAM.
• Can be run at very high resolutions.
• Less than 2km!

4
Why?

• MCS systems are especially difficult events to
  forecast in Western North Carolina due to the
  Appalachian Mountains.
• Current forecast models have difficulty creating
  high resolution products while acknowledging this
  complex terrain.
• The WRF has the capability to create high
  resolution products while recognizing unique land
  features.

5
Criteria for Selecting Events to Study

• Crossing.
• MCS crosses over the mountains while retaining
  most of its energy.
• Dissipating.
• MCS dissipates and/or loses most of its energy
  upon reaching the mountains.

6
First Event

• April 22nd, 2005.
• Involves two waves of severe weather.
• First wave crosses, second dissipates.
• One tornado report in Pickens County, SC.

7
Second Event

• May 19th, 2005.
• MCS crosses over the mountains but is severely
  weakened.
• One confirmed F1 tornado in Smyth County, VA.

8
Third Event

• July 27th, 2005.
• Very broad and linear line of convection.
• As the system advances upon the Appalachian
  Mountains it quickly dissipates.

9
First Event Radar Imagery

• So far we have been able to run the first event
  (April 22, 2005) in the WRF model.
• All radar images are 0.5 degree base
  reflectivity.
• Radar data is courtesy of the NCDC.
• Radar images were made using StormLab with help
  from Chad Hutchins.

10
April 22 at 18Z. First Wave.
11
April 22 at 19Z. First Wave.
12
April 22 at 20Z. First Wave.
13
April 23 at 01Z. Second Wave.
14
April 23 at 04Z. Second Wave.
15
April 23 at 05Z. Second Wave.
16
First Event WRF Data

• This run of WRF was initialized at 06Z April 22
  (the day of the event).
• Map of 500mb vorticity.
• Vorticity is a good indicator of convection.

17
April 22 at 18Z. First Wave
18
April 22 at 19Z. First Wave
19
April 22 at 20Z. First Wave
20
April 23 at 01Z. Second Wave
21
April 23 at 04Z. Second Wave
22
April 23 at 05Z. Second Wave
23
Radar Data Vs. WRF Forecast

• Observations with the first wave.
• WRF appears slow to bring in convection with the
  first wave.
• Leaves residual bull's-eyes of vorticity
  afterward.
• Observations with the second wave.
• WRF initially looks mostly on target with onset
  of front at 01z.
• However, it does not even show any showers
  crossing at all at 04Z and 05Z.

24
So How Did The WRF Do?

• WRF at first glance has problems with onset of
  prefrontal convective precipitation.
• However, it was still surprisingly accurate.
• The model had problems bringing the actual
  frontal precipitation across the mountains.
• Theories or hypotheses?
• Did the model handle this well because of the
  synoptic features influencing the event?
• Changes to make to WRF runs?
• Initialization time appears to be important.
• 00Z initialization was very slow.
• 12Z initialization missed a lot of the convection
  (ironic).