Page 3 of 4

Re: How to edit GFS input gribs?

Posted: Wed Sep 25, 2013 6:15 pm
by meteoadriatic
Excellent! I'm going to try, this is even better (this way no negative values are written).

Re: How to edit GFS input gribs?

Posted: Wed Sep 25, 2013 6:20 pm
by Antonix
yes!
wish to impose any condition you can do it using parentheses.

http://nco.sourceforge.net/nco.html

Re: How to edit GFS input gribs?

Posted: Wed Sep 25, 2013 6:24 pm
by meteoadriatic
It works flawlessly! OK next... find out some snowy gribs from past and test out change :)

We will need to create little shell script for doing this command and modify strc scripts in a way that this modification is done after metgrid and before real. That will be easy. And it is done, much easier than I expected :mrgreen:

Re: How to edit GFS input gribs?

Posted: Wed Sep 25, 2013 7:41 pm
by Antonix
ok! I have grib GFS of a very snowy period (December 2012/ genn 2013). we want to try???
you have dropbox or similar (or this forum)
in changes in the code STRC I'm no very expert!

this is what you want to do?!
1)ems_prep xxx
2) script shell for modify .nc
3)ems_run

Re: How to edit GFS input gribs?

Posted: Wed Sep 25, 2013 7:58 pm
by meteoadriatic
Antonix wrote:ok! I have grib GFS of a very snowy period (December 2012/ genn 2013). we want to try???
you have dropbox or similar (or this forum)
It is easy to find (even strc servers contain archives). Currently I need to do some stuff for tomorrow busy work day so I can't test right now.
in changes in the code STRC I'm no very expert!
I'll help with that, this will be trivial change.
this is what you want to do?!
1)ems_prep xxx
2) script shell for modify .nc
3)ems_run
Yes that is what I thought.

First met_....nc file found in $EMS/runs/<domainname>/wpsprd should be treated with such command before ems_run (real) executes. You need to do it twice, once for SNOW variable and once for SNOWH. For SNOW, delete everything below approx. 10 kg/m2, and for SNOWH delete below approx 0.1 m. Those should be OK values to start testing with. We can do something better later (for example gradually change values or something like that).

But now I have little doubts about where to do modifications. I'm not 100% sure that changing only analyze time is enough. I don't know if WRF updates ground fields with forecast input files or not. If it updates (I don't think so but we need to check it) then it is not enough to modify only analyze .nc file but either all of them (after metgrid), or modify wrfinput file (after real). However, one quick test with both approaches will reveal if there is any difference.

Re: How to edit GFS input gribs?

Posted: Thu Sep 26, 2013 3:21 pm
by meteoadriatic
Tried this, but ems_run does not want to run, it exits with error if I put there modified nc files. I'm trying to find out why. Do you have same experience?

If I only put modified .nc for domain 2, and leave unmodified files for domain 1, then it works. Strange, I need to see what's going on.

EDIT: OK, it is something with AUTO_S timestep calculation in strc scripts that doesn't like modified file. No idea why but now it is not high priority bug. It works with manual timestep selection and (probably will) with adaptive ts. I'm running now first simulation with modification.

Re: How to edit GFS input gribs?

Posted: Thu Sep 26, 2013 4:56 pm
by Antonix
in this situation are more important this setting

input_from_file = T, T
fine_input_stream = 0, 2
surface_input_source = 1
this is a default configuration

I use
T,T
2,2
3

in this way each domain uses its values ​​of soil, landuse etc etc etc..
each domain uses the GFS data and not data "seasonal"

Re: How to edit GFS input gribs?

Posted: Sun Oct 06, 2013 11:57 pm
by johnbasham
Could this same type of programming be used to insert additional surface data into the original GFS GRIB files? For instance, I have a script that scrapes real-time weather data from privately owned weather stations (meso-nets) across North America.

If I did use this to modify and or insert additional surface data into the GFS initial GRIB files would it make a marked difference in the near-term forecast data output by the NEWR EMS?

Re: How to edit GFS input gribs?

Posted: Mon Oct 07, 2013 6:01 am
by meteoadriatic
Yes, theoretically it would be possible. But, for that matter, there is much better solution:

http://www.mmm.ucar.edu/wrf/users/docs/ ... _chap7.htm

Re: How to edit GFS input gribs?

Posted: Sun Nov 10, 2013 8:59 am
by Daniel
@ meteoadriatic: Did you see any improvements from your tests?

Regards

Daniel