Future Forecast Settings

Discuss advection schemes, timestepping, pressure gradients, coordinates, etc.

Future Forecast Settings

Postby Juarez.VO » Tue Apr 09, 2013 1:40 pm

Hello everyone.

First of all, I'm new to WRF and to Unix, but I've succeded on compiling WRF (now I set it to smpar), WPS and WRFDA using gfortran on a OpenSuse machine, and I've also succeded on running a few test cases, like the Katrina one.

I've downloaded 9 GFS Analysis Only files, from 2013-04-07_00 to 2013-04-09_00, I can run the model perfectly for the same period.

My question: how can I make a forecast beyond the last GFS file date?

For exemple: my start date is set to 2013-04-08-00 and I want it to end not on 2013-04-09_00, but on 2013-04-10_00.
I've setted the namelist.input to run for 2 days and it runs out of out of valid boundary conditions in file wrfbdy_d01. I've also tryed a restart run, without succes.

Any help will be appreciated.

My namelist.input with a few modifications that I try to solve the problem:
&time_control
run_days = 3,
run_hours = 0,
run_minutes = 0,
run_seconds = 0,
start_year = 2013,
start_month = 04,
start_day = 07,
start_hour = 00,
start_minute = 00,
start_second = 00,
end_year = 2013,
end_month = 04,
end_day = 09,
end_hour = 00,
end_minute = 00,
end_second = 00,
interval_seconds = 21600
input_from_file = .true.,
history_interval = 180,
frames_per_outfile = 1000,
restart = .false.,
restart_interval = 1440,
io_form_history = 2
io_form_restart = 2
io_form_input = 2
io_form_boundary = 2
auxinput4_interval = 60
io_form_auxinput4 = 2
diag_print = 2
auxhist2_outname = "rainfall_d"
debug_level = 0
Juarez.VO
 
Posts: 5
Joined: Tue Apr 09, 2013 11:08 am

Re: Future Forecast Settings

Postby Phillip » Tue Apr 16, 2013 6:52 am

If you don't run the model globally your domain has boundaries which are affected by the conditions outside of the domain. As a result, you need information for either the boundary (which is contained in the boundary files) or you have to simulate the outside. The only domain which does not require boundary files is one covering the whole earth, since it is periodic.

In other words: If you want to run the model without initial data for all time steps you have to do a global run. Inside the global simulation you can nest smaller domains for downscaling. Initializing a global run only requires one time step of global input data.

Where is your domain located and how large is it?

Check this page for information on how to start a global run:
http://www.mmm.ucar.edu/wrf/users/docs/ ... htm#global
It's the chapter of the WRF manual which explains how to start a global run.
Phillip
 
Posts: 46
Joined: Thu Jul 19, 2012 3:29 am

Re: Future Forecast Settings

Postby Juarez.VO » Tue Apr 16, 2013 9:24 am

Thank you very much, Phillip, your answer was just what I was looking for.
Juarez.VO
 
Posts: 5
Joined: Tue Apr 09, 2013 11:08 am

Re: Future Forecast Settings

Postby Phillip » Tue Apr 16, 2013 11:58 am

Glad I could help :)
Phillip
 
Posts: 46
Joined: Thu Jul 19, 2012 3:29 am

Re: Future Forecast Settings

Postby bluefish1188 » Thu May 02, 2013 11:14 pm

Phillip wrote:If you don't run the model globally your domain has boundaries which are affected by the conditions outside of the domain. As a result, you need information for either the boundary (which is contained in the boundary files) or you have to simulate the outside. The only domain which does not require boundary files is one covering the whole earth, since it is periodic.

In other words: If you want to run the model without initial data for all time steps you have to do a global run. Inside the global simulation you can nest smaller domains for downscaling. Initializing a global run only requires one time step of global input data.

Where is your domain located and how large is it?

Check this page for information on how to start a global run:
http://www.mmm.ucar.edu/wrf/users/docs/ ... htm#global
It's the chapter of the WRF manual which explains how to start a global run.


Hi,I have a question about the forecast time.
I think that forecast in WRF model should be started from one known time peroid . for example, I have the dataset(gfs ) donwload from website , the start time is 2009-1-1_00 and the end time is 2009-1-2_00. I want to forecast 2009-1-3 ,1-4.......etc. But I expected the forecast happen in the region , not global .

in other word ,I want to know whether WRF model needs to setting the range of time exceeding the date of input data.
bluefish1188
 
Posts: 4
Joined: Thu May 02, 2013 5:57 am

Re: Future Forecast Settings

Postby AchternDiek » Mon May 06, 2013 3:29 am

bluefish, I am not sure if I understood your question correctly.
But, as Phillip wrote, for regional forecasts you need to have valid boundary conditions (e.g. GFS data) for the whole simulation period. Otherwise, what incoming atmospheric flow should be modeled at the region's boundaries? How should the model know about an incoming low pressure system, for example.

--AchternDiek
AchternDiek
 
Posts: 147
Joined: Wed Jun 06, 2012 6:10 am
Location: Norway

Re: Future Forecast Settings

Postby bluefish1188 » Tue May 07, 2013 10:24 pm

thanks!I understand your word. the WRF forecast run setting must need the boundary condition. I ran forecast case with the RAMS model before, in this model the time setting is started from one time point. I just want to learn know what are different between the WRF model and RAMS model on forecasting.
bluefish1188
 
Posts: 4
Joined: Thu May 02, 2013 5:57 am

Q: Future Forecast Settings

Postby anu » Mon Jun 17, 2013 2:34 pm

Hi, i want to do a future run of wrf.
i have gfs data till 2013-06-17:00 i want to forecast ,,,
can someone help me with sample namelist.wps and namelist.input in order to understand how to provide start date and end date and global boundary and nest domain in order to run for forecasting.
anu
 
Posts: 13
Joined: Wed Aug 29, 2012 12:40 am

Re: Future Forecast Settings

Postby Phillip » Thu Jul 25, 2013 10:52 am

Do you need help with the namelist.wps or namelist.input?

The first one is required for the pre processing which creates your initial fields and boundary condition files from model data.
The second one is for the model itself.

In both of them setting the time frame is pretty straight forward, just replace date and time. Have you already set your domain? (using geogrid.exe and so on?)
Phillip
 
Posts: 46
Joined: Thu Jul 19, 2012 3:29 am

Re: Future Forecast Settings

Postby anu » Fri Sep 13, 2013 2:46 am

thanks ..
but yes i hv figured out now there is no forecasting using a prior date ... its only .. using already forecasted data by GFS model and providing initial conditions to WRF model....
anu
 
Posts: 13
Joined: Wed Aug 29, 2012 12:40 am

Next

Return to Dynamics and Numerics

Who is online

Users browsing this forum: No registered users and 2 guests