|
http://www.atmos.washington.edu/academics/classes/2013Q1/380/HW4.html Due Thursday Feb 7 |
|
In this exercise you will learn about the purpose of running multiple ensemble members to investigate error in a forecast associated with varying model physics. We will be using WRF-ARW for this exercise to simulate Hurricane Katrina. We won't undertake any validation yet. Instead, we will compare ensemble members. The grid resolution is 15km in the horizontal. The boundary conditions are derived from "analysis" (The GFS forecast model 1 hour after initialization). This run is designed after the WRF Hurricane Katrina Tutorial. I don't recommend that you attempt to follow the instructions in the tutorial at this time. The link is provided for those who want to see what I did behind the scenes. Summary: I. Copy the files you need for this exercise. Edit the namelist for your to vary the convection. II. Run WRF III. Analyze output in MATLAB and write up answers to a few questions (a couple of hours max) Thanks go to Luke Madaus and Bonnie Brown for answer my questions about WRF. I. Copy files and make links Make a directory for this exercise in our class directory area, call it WRFV3. CAUTION: look for dots in the blue text below cd /home/disk/p/atms380/$LOGNAME tar xvf ../WRFV3_HW4.tar cd WRFV3 Look around this dirctory a bit. It is exactly as you would download from NCAR, except I have compiled the code already. In other words, I have gone through the steps I described in lecture on 1/29. We are going to do a "real" run with WRF, which means the run uses real input rather than idealized data. So you will go to a directory set up for real runs and link to data that I prepared aleady. cd test/em_real ln -sf /home/disk/p/atms380/specialinputs/wrfbdy_d01 . ln -sf /home/disk/p/atms380/specialinputs/wrfinput_d01 . II. Edit the namelist and run WRF Edit the namelist.input for your particular case: Nate change cu_physics to 2 Steven change cu_physics to 3 Shane change cu_physics to 4 Andrew change cu_physics to 5 Sebastian change cu_physics to 6 Jessica change sf_sfclay_physics and bl_pbl_physics to 2 Adam change sf_sfclay_physics and bl_pbl_physics to 3 David change sf_sfclay_physics and bl_pbl_physics to 4 Ali change sf_sfclay_physics and bl_pbl_physics to 5 Note that there are three numbers after the equal sign. You only need to change the first one. The others are only used for nested grid cases. When done editting the namelist submit the job to the queue. Also for sanity check, verify it is in the queue. qsub run-wrf.csh qstat qstat -u "*" the latter lets you see all the jobs in the queue. If for some reason you wish to kill your job. qdel xxxxx fill in the x's with the job-ID and this will cancel your job Wait about 20 min and when qstat indicates you job is no longer in the queue, check to see that you have an output file that is ABOUT this size (hint use ls -l wrfout*) -rw-r--r-- 1 bitz atgstaff 690332288 Jan 30 08:42 wrfout_d01_2005-08-28_00:00:00 When you are sure your job is done, ask Cecilia for a run number. Then move the output file to a name with "_runX.nc" on the end, where X is your run number. For example if Cecilia gives you #1, then mv wrfout_d01_2005-08-28_00:00:00 wrfout_d01_2005-08-28_00:00:00_run1.nc Copy your run to a group directory for everyone to access. Use the run number I give you, not necessarily run1 as shown. cp wrfout_d01_2005-08-28_00:00:00_run1.nc /home/disk/p/atms380/wrf_katrina Go to that directory and and run a script with a program called "ncl" to compute sea level pressure for your run. (FYI, ncl = ncar command language, which is a powerful piece of software akin to matlab for number crunching and making images. It is not nearly as easy to use, but it is free.) You will have to customize the getslp.ncl script so it reads in your file. Start by copying it to a new name just for your run. Edit the script and then run ncl as shown. Use the run number I give you, not necessarily run1 as shown. cd /home/disk/p/atms380/wrf_katrina cp getslp.nc getslp_run1.ncl (EDIT THE getslp_run1.ncl script!!!) ncl getspl_run1.nc III. Analyze WRF - turn in about a couple of pages, plus figures. The MATLAB scripts for this exercise and are in a directory named /home/disk/p/atms380/wrf_katrina . Copy them wherever you wish. a) Run ex4_a.m and examine the results for my run while waiting for the rest of the runs to finish. Make movies if you want. There is nothing to write up yet. If you make a movie and save it as a gif file, such as Katrina_run1.gif, copy the file to your directory ~/public_html and point a browser to the file at http://www.atmos.uw.edu/~yourloginname/Katrina_run1.gif Then send the link to your siblings and tell them science is fun. b) Run ex4_b.m in matlab. This script plots the cumulative rainfall either from convective clouds only or all types of clouds. Run the script once as it is. Then edit "thevar" to plot the other type of rainfall. Discuss the difference. Discuss spurious behavior near the edges when thevar='RAINC'. Examine a few different runs. How much does the maximum cumulative rainfall change among the runs? Include figures to augment your discussion. c) Run ex4_c.m This script tracks the hurricane by plotting the minimum in sea level pressure in the domain. Include as many of the runs as possible at once with minor edits. What do you see? Does it surprise you considering we only varied the convection and planetary boundary layer physics? What kind of ensemble is this (of the three kinds discussed in class)? d) Run ex4_a.m again but for some of the runs done by the class. Be sure to look at any of the runs that are especially strange. Discuss the behavior of the strange one with regard to its movement relative to the domain edge. If you are eager to know more about WRF output, edit ex4_a.m to contour in color the sea level pressure instead of wind speed. Or try to plot some other variable. You can find a list of the other variables in the file named /home/disk/p/atms380/wrf_katrina/header.
|
| Return to Homepage |