Long-term Forcast

All issues/questions about EMS v3.4 package, please ask here.
Post Reply
PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Long-term Forcast

Post by PedroKannenberg » Mon May 11, 2015 11:55 pm

Hi, I'm relatively new in the usage of EMS and I would like to know if there's a way to make long-term forecasts (more than 16 days). Do I need to have all the gfs files from the servers to run ems_run? My objective would be a 30 day forecast of a relatively small region. I know that the forecast would lose some of it's accuracy but nevertheless, I would like to know if there's a way of running it.
Thanks!
All the best,
Pedro Kannenberg
ppbr41@hotmail.com

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Long-term Forcast

Post by meteoadriatic » Tue May 12, 2015 6:33 am

Hello,

You should use something like CFS for your input:
http://nomads.ncep.noaa.gov/pub/data/nccf/com/cfs/prod/

Also, take a look into wrfems user's manual at option nudging. It will be very useful for such simulations.

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Tue May 12, 2015 10:22 am

Ok, thank you very much for the quick response and thank you for the advice, that's precisely what I needed.
Thanks!
Regards,
Pedro Kannenberg

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Tue May 12, 2015 5:28 pm

I'm sorry for my ignorance but, I took a look on the cfs files and realized that those files go only as far as the present day. How do you use them to predict forward in time? I always downloaded the gfs files of some days forward (e.g. perl ems_prep --dset gfs --length 68) and then used the ems_run routine (perl ems_prep) to see the results. Is there another way of doing it?

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Long-term Forcast

Post by meteoadriatic » Tue May 12, 2015 10:32 pm

Maybe you look at wrong place.

http://nomads.ncep.noaa.gov/pub/data/nc ... y_grib_01/

I believe, files that starts with pgrbf are those you need. They go from analyse up to December, in 6 hours step.

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Tue May 12, 2015 10:49 pm

I see, I have indeed looked in the wrong place, thanks a lot once more.
All the best,
Pedro Kannenberg

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Thu May 14, 2015 3:42 pm

Hi there, I'm trying to set my cfs_gribinfo.conf file and I'm having difficulties with the server-method, do you know how it should be? Do I need to set another Vtable or can I use another existing one?
Thanks!

meteoadriatic
Posts: 1567
Joined: Wed Aug 19, 2009 10:05 am

Re: Long-term Forcast

Post by meteoadriatic » Thu May 14, 2015 4:39 pm

Hi,

I tried to investigate a bit and it seems it is not trivial to run WRF with cfs...
http://computing.io/wp/2013/08/wrf-with-cfsr-data/

There you will find information about vtables but also other important things you will probably need to address somehow, which seems not easy inside wrfems environment (you will need to run ungrib twice, once for surface data, once for upper air):
http://forum.wrfforum.com/viewtopic.php?f=6&t=5627

Maybe.. don't know really, never did something like that.

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Thu May 14, 2015 5:12 pm

The first link you sent was about cfsr, isn't that data type for reanalysis? I thought CFS was different from the CFSR input. I initially thought that CFS was the same as CFSR and then saw on the wrfems guide and the currently existing cfsrgribinfo.conf information about it and everthing I saw about it says that cfsr is used for reanalysis. I coudn't download correctly the CFS files with ems_prep though, mostly because of my server-method which I didn't manage to write.

PedroKannenberg
Posts: 8
Joined: Mon May 11, 2015 11:43 pm

Re: Long-term Forcast

Post by PedroKannenberg » Fri May 15, 2015 12:37 am

Is there any other way of making such long forecasts?

Post Reply