1. Supercell storms: In-class demo and Experiment 3
ATM 419/563
Spring 2017
Fovell

2. Goals
Start idealized WRF 3D run to simulate a splitting supercell thunderstorm (in-class demo)
Review supercell storm structure and behavior (while model runs)
Analyze demo simulation
Investigate sensitivity to microphysics
Employ GrADS ensemble dimension, and compute ensemble mean and standard deviation
Submit WRF jobs to the Snow batch system

3. In-class portion

4. Setup Create a folder called SUPERCELL in your lab space
Copy /network/rit/lab/atm419lab/SUPERCELL/SETUP.tar there and unpack
Files included: make_all_links.csh, namelist.input, input_sounding, read_wrfinput.py, submit_wrf, control_file.z
Pre-configured for a 2-h run in a 168km by 168km domain at 2km horizontal resolution
Initial sounding derived from Weisman and Klemp (1982)
“quarter-circle” wind profile favors storm splitting and right-mover
Purdue Lin microphysics (mp_physics = 2)
Time step 12 seconds
History output every 10 min
Computes some AFWA diagnostic fields (&afwa section)
read_wrfinput.py
WK sounding was slightly modified from original

5. Boundary layer mixing
ratio is 14 g/kg in
input_sounding
Weisman and Klemp (1982)

6. Hodograph (see input_sounding)
2 km AGL
7 km AGL
surface

7. Using the Snow batch system
Look at the file submit_wrf:
#!/bin/bash
# Job name:
#SBATCH --job-name=ATM419
#SBATCH -n 6
#SBATCH -N 1
#SBATCH --mem-per-cpu=7G
#SBATCH -p snow
#SBATCH -o sbatch.out
#SBATCH -e sbatch.err.out
[…]
echo "running WRF"
srun -N 1 -n 6 -o wrf.srun.out ./wrf.exe
Change to your name
Requests 6 cpus on 1 node
Where run time errors might be reported
This runs the model.
The –N and –n values must match
those provided at the top

8. Initialize model Allocate resources
csh make_all_links.csh
srun ideal.exe
This MPI version does not write to screen anymore
To see if the job worked: type trsl, which stands for “tail –f rsl.out.0000”. Type CTRL-C to break free.
What is the lowest scalar model level height?
python read_wrfinput.py wrfinput_d01
Submit your batch job:
$ sbatch –p snow submit_wrf
Check on your job
$ squeue –u yournetid
Lowest scalar height: 285 m. Not uncommon for cloud model simulations. Not even trying to capture PBL properly.

9. As the model runs…
The model writes output to four kinds of files (in addition to wrfout…)
rsl.out.XXXX (numbered 0000 to 1-n, n =number of nodes requested)
rsl.error.XXXX (numered as above)
sbatch.out (timing info when job completed)
sbatch.err.out (hopefully it is empty!)
As the model runs, you can “watch” it with
$ trsl
[This is an alias for tail -f rsl.out.0000]
[type ‘CTRL-c’ to quit/escape from this]
Look for SUCCESS COMPLETE WRF as usual

10. Supercell structure and behavior

11. Supercells
Characterized by rotating updrafts (vertical velocity and vertical vorticity co-located)
Mesocyclone and “hook echo”
Develop in environments with large vertical wind shear
Evolve from ordinary cells by “splitting” into “left-movers” and “right-movers”
Clockwise directional vertical wind shear favors right-movers

12. Split L mover weaker R mover turns right R mover hook Purple = hail
24apr2006_split_dfx_n0r01.GIF
Split
L mover weaker
R mover turns right
R mover hook
Purple = hail

13. Splitting storms
Wilhelmson and Klemp (1981)

14. Shear-induced horizontal vorticity

15. Dynamics of storm splitting
• Lifting of vortex tubes by updraft
creates counter-rotating vortices
(vertical vorticity)
• Low perturbation pressure develops
that establishes new updrafts on
flanks of original storm
• New cells are rotational: CCW or CW
• Old cell dies, giving appearance of
split
• Actually, precipitation does NOT
cause the updraft split

16. Original updraft

17. Counter-rotating vortices

18. New storms form on flanks

19. Old cell decays

20. reflectivity
> 65+ dBZ
(hail or airborne debris)
hook echo

21. radial velocity

22. Supercell conceptual
model

23. Storm-scale fronts
T = tornado location

24. mesocyclone
meso = Gk., middle

25. Updraft velocities can easily
exceed 100 mph

26. RFD = rear flank downdraft
FFD = forward flank downdraft

27. warm, moist air
lifted into updraft

28. Supercell storm structure
Plan view at surface
UD = updraft
FFD = forward flank downdraft
RFD = rear flank downdraft
Gust front
Hook echo
Dot ~ tornado location
Lemon and Doswell (1979)

29. {Do Supercell in-class demo here}

30. Experiment 3

31. EXP03 overview
Make 5 different simulations with 5 different microphysics schemes
Do not select schemes 1, 3, 4, 5, 11, 12 or 14, as they do not produce radar reflectivity fields and/or have graupel as a species
Do not select scheme 2, as we did that in class
Do not choose bin schemes 30 or 32
Use the default time step of 12 sec
srun ideal.exe as usual. Check tail of rsl.out.0000 file.
WRF simulations have to be submitted to the Snow batch queue, using the submit_wrf script
Given our setup, you can only run one job at a time
Unpack the 5 simulations using control_file.z, using a naming convention like mpXX, where XX = mp_physics
Note that some schemes create QHAIL but control_file.z will ignore this field. The ensemble members need identical CTL files

32. EXP03 tasks You have created a small WRF microphysics ensemble
Write a GrADS CTL file for the ensemble
Use the SQUALL demo’s mp_ensemble.ctl for guidance
Run GrADS and open your ensemble CTL file
Write a GrADS script to compute the mean and standard deviation of w_up_max at the final time (2 hours)
Std deviation shaded and mean field contoured and superimposed.
Use squall_ensemble.gs for guidance.
Make a plot, and label axes
Send to me:
A list of which microphysics schemes you selected
A PNG plot (or screenshot) of your w_up_max field
Your GrADS script
Your ensemble CTL file
One additional plot, properly labeled, of anything you want. (Make sure I know what I am looking at.)