Restart problem

Forum dedicated to older versions of EMS package (WRFEMS v3.2, v3.1 or older). Support is user-to-user based, so please help others if you can.
Post Reply
Posts: 260
Joined: Fri Oct 16, 2009 8:53 am

Restart problem

Post by Antonix » Thu Oct 18, 2012 4:51 pm


I have to make a simulation of three months.
but on the same machine running an operating model (ROMS model + SWAN model).

So I have the need to run the WRF more run.

I would like to make the restart but I always get this error:

EMS ERROR: Simulation Length Issue

The requested ending date / time of your simulation extends
beyond the available data:

Requested Start: Thu Oct 18 09:00:00 2012 UTC
Requested End: Thu Oct 18 15:00:00 2012 UTC
Last Available: Fri Oct 19 03:00:00 UTC 2012

Requested Change the length of your forecast or rerun
ems_prep to create additional boundary condition files.

whatever the command that use.
for example :
ems_prep --date 20120101 --cycle 00 --length 240 --nogribdel --local --dset fnl
./ems_run --length 24


./ems_run --restart 2012-01-02_00:00:00 -length 24h

it may be a bug, I have something wrong??
you have any idea to do a restart??
best regards

Posts: 176
Joined: Sun Jun 24, 2012 8:42 pm
Location: Los Angeles, CA, USA

Re: Restart problem

Post by pattim » Fri Dec 28, 2012 5:47 pm

Hi Antonix: It says it's not finding enough boundary condition (BC) files for the forecast. I'm trying to figure out how to forecast into the future (there aren't any BC's of the future). It seems to be kind of a fundamental problem with weather forecasting. Any time you try to forecast into the future with a Local Area Model like WRF, you don't have BC's for the run because the future has not happened yet - but the solver always needs BC's to do it's mathematics. I am not sure how people actually circumvent this issue in practice. I believe it may be a switch in EMS somewhere to tell it to use the same (current) BC's for all future times. (Or you could use last year's BC's on the day of interest.) But in either case, errors from the BC's propagate at a finite speed into your domain - I *think* the fix is to make sure your grid boundaries are far enough away from your region of interest so that the BC errors do not reach it in the time of your forecast. (You always have to wait *some* amount of time for dynamical consistency to be achieved, however, usually several hours or so... at least according to Warner.)

There are no BC's on a Global WRF simulation, so you could do a coarse Global WRF - and then if you nest-down a region of interest to finer grid scale, it will obtain its BC's from the Global WRF simulation and not request current BC's for the nest.

Other suggestions/explanations greatly appreciated!! - - - Especially how to set up EMS to do these things (switches, options, conf-file-edits, and such).

Happy Holidays!

Patricia 8-)

Posts: 176
Joined: Sun Jun 24, 2012 8:42 pm
Location: Los Angeles, CA, USA

Re: Restart problem

Post by pattim » Thu Jan 03, 2013 5:40 am

I just tried doing a global WRF run and it only needed a single grib file for the initial conditions. I am not sure how to deal with a local area - it will always need some boundary conditions. Maybe there is a switch you can set to hold the boundary conditions constant? That would work if you put your boundaries thousands of km away from your region of interest and did not forecast for too long... Maybe someone will tell us how to do this with EMS.

Posts: 1
Joined: Thu Jan 31, 2013 11:46 am

Re: Restart problem

Post by allen05 » Thu Jan 31, 2013 11:47 am

Hello everybody,

I'd like to know if it's possible to include data from
my weather station to initialize my wrf model.

Thank you!


Post Reply