[Wrf-users] properly removing topography from a restart simulation?

[Russ Schumacher]

Marc and all,

I'm not sure there's a way to do this reliably from a restart run, because remember that the WRF vertical coordinates are terrain following!  The steps you describe with running metgrid and real.exe all sound reasonable for initializing a run -- if you alter terrain in the geo_em file, then the metgrid and real.exe steps will put the terrain-following coordinates in the right places based on the surface pressure and elevation, and will hydrostatically balance the initial state to your specified terrain.  There are issues with what to "fill in" the atmosphere with in the places that are no longer below ground, but if you're using NCEP grids as initial conditions then they do have information on pressure levels below ground (which are calculated from the US Standard Atmosphere) and in my experience this won't cause too many problems after the standard model spin-up time of a few hours.  

However, if you're going from a restart file then the surface pressure and corresponding vertical coordinates are going to have values as if there's a mountain there, like there was in your original run.    And all of the other state variables are going to have values that are related to these.  Presumably the model atmosphere eventually will adjust to the HGT field being zero, but I'm not sure exactly how that variable then interacts with everything else, how long it'll take for the surface pressure (which isn't really the surface pressure anymore) to adjust, etc.  So I'd be very cautious about trying to alter the terrain in a restart run, as you ideally would want to rebalance the initial state to account for the new terrain, but in which case it might just be easier to initialize separate runs.

Another option you could try is to use ndown.exe on your original run, applying met_em files with and without the terrain modifications (appropriate initial surface pressures and vertical coordinate structures should be applied during the process of running ndown), and then start 2 new simulations from those initial conditions, which might work if you can accept a slight change to your original domain configuration.  (You might even be able to effectively set up the same domain as your original when running through ndown, though I've never tried that.)

Russ




On Jul 29, 2015, at 7:45 AM, "Marcella, Marc" <MMarcella at AIR-WORLDWIDE.COM> wrote:

> Hi wrfusers,
> 
> I am trying to remove topography over Taiwan in a WRF simulation where I restart from a prior run to simulate the effects on a typhoon traversing the region.  The procedure I follow is to remove the topography (HGT_M, VAR_SSO, and SLOPECAT variables) in geogrid, re-run metgrid.exe, then re-run real.exe.  I checked the geo.nc, met.nc files, and the real input files---all show the topography variables over the region as zeroed out.  I then am sure to remove the topography from the wrfrst.nc file that the simulation is restarting from.  
> 
> When I run the model however, the results do look slightly different than a simulation with topography but not significant given the drastic change (from having topo go from 3km to 0km). Similarly, the results still show a barrier of sorts when the wind/rainfield comes ashore.  When I check the restart files created after the fact, I see that the topography is still 0 but the surface pressure over the island appears as if mountains are there as it is very low (~800mb) as is the sfc pressure in the met_em.nc files.  My question is, am I properly removing the topography? Or are there other variables that also must be zeroed out to mimic a 0km island?  And, is there some other definitive check I can perform to ensure that this simulation is actually not seeing the topography?
> 
> Any help would be greatly appreciated. 
> 
> Thanks,
> Marc
> _______________________________________________
> Wrf-users mailing list
> Wrf-users at ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/wrf-users



--
Russ S. Schumacher
Assistant Professor
Department of Atmospheric Science
Colorado State University
e-mail: russ.schumacher at colostate.edu
phone: 970.491.8084
web: http://www.atmos.colostate.edu/faculty/schumacher.php






-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/wrf-users/attachments/20150729/037046dc/attachment-0001.html