From Sai Ming:
I find that the NCEP CFS produces daily forecast SST in GRIB format. Do you think it is appropriate to put these forecast SST values into the GSM run?
From Sai Ming:
As you may know, the HKO currently downloads boundary conditions from ECPC to run our regional model (RSM). The names of the files are r_sfcG.fhhhh.asc.gz and s_r_sigG.fhhhh.asc.gz. The hhhh is forecast time indicator. I think these files are just extracts of the GSM output converted to a certain format. Do you know if there are any tools to generate such files from GSM output?
see rsm/src/rinpg/rinpg.F
So you want to get operational analysis from NCEP, run GSM, and run RSM with its output. Correct? It is not necessary to extract GSM data and reformat it for RSM if you do everything at your institution. RSM is able to read GSM output directly.
A question from user sreejith
---
I would like to run the model with persistant SST file. ( SST anomaly of march 2006 add to the climatological SST file for the months april to dec). How can I replace the climatological SST file with my sst file?. OR any other method is there for persistant SST run. My file is in binary.
SST files are specified in the script on two lines:
SST_ANL_T0=
SST_ANL=
G-RSM comes with climatological SST files in GRIB format: libs/etc/clim.sst.grib
You need to convert your binary SST data to GRIB using a tool such as libs/etc/ieee2grb_sst . Also make sure you set the date for each record in your SST grib file correctly. For this, libs/etc/incdte and libs/etc/force_grib_date_mon are useful.
Hideki
From hko:
2. There are a number of climaology files, e.g. ice, used in running GSM. If the GSM is put into real time operation, would you recommend to use other files (maybe more updated and more appropriate) instead of the original climatology ones? For example, I can use the SST forecast produced by NCEP CFS instead of the SST climatology.
If you provide climatology (without providing analysis), the program first checks whether it is annual, seasonal or monthly climatology, and interpolate them to the date corresponding to the forecast hour. There is no choice for daily climatology. I do not think you can use climatological file as a replacement for daily analysis or forecast files. If you want to use CFS produced SST, specify them as a SST analysis file.
Hi everyone.
I am trying to run the NCEP GSM using initial conditions that I got from nomad3.ncep.noaa.gov. My run should start at 00z01April2000, therefore I've downloaded the following files: sig.anl.2000040100.ieee and sfc.anl.2000040100.ieee. The SST data is also for the year 2000.
However, the model can't go beyond the cnv4dig script, from which I am getting the following message:
date record of sig and sfc does not match
sigma: 0/04/01/00/0
sfc : 2000/04/01/00/0
Both files are for 00Z 01 Apr 2000. But it seem like the year in the sigma file was written as "00", whereas in the surface file it was "2000". This is causing the model to abort.
Does anyone know how to solve this problem?
I'd appreciate any input. Thanks in advance.
Hi - I just tested GSM with cases script starting from 20000401 with r2 files (sig.anl.2000040100.ieee and sfc.anl.2000040100.ieee) which are on our ECPC data server, and the model worked fine. I looked at the log but cnv4dig was not used. Can you tell me a little more about how your experiment was set up? Hideki
Hi. I ran a few NCEP GSM simulations many year ago, for which I utilized snow cover (SNOWC) data as boundary condition. The snow cover data was then downloaded from the NCEP data server. All my runs were for cases between 1987 and 1997.
Now I find myself having to run the GSM for more recent years. I have noticed, however, that the NCEP does not provide snow cover in their data server (nomad3.ncep.noaa.gov) anymore but snow water equivalent instead (WEASD).
I was wondering if anyone in here has tried to use WEASD as boundary condition in their simulations. If so, did you have to modify the code at all or the model can automatically detect the type of information.
I'd appreciate you help.
Thank in advance.
Hi,
I can find SNOWC on NOAA's ftp site:
ftp://ftpprd.ncep.noaa.gov/pub/data/nccf/com/gfs/prod/gdas.YYYYmmdd/gdas1.tTTz.snogrb
The second record of the grib file (snogrb) is SNOWC. I believe you can use wgrib to extract the record.
I haven't tried to run GSM with WEASD because I learned that snow cover can be predicted by the model. So I just let the model work it out.
Cheers,
Sai Ming
Hi,
Can I use the NCEP/DOE R2 data to initialize the GSM? If yes, what do I need to modify, e.g. the gsm script?
Sai Ming
Yes, you can use R2 data.
Please download R2 sigma/sfc files from sftp://tsuzumi.ucsd.edu (see ECPC%20data%20server.html )
And edit your INITIAL_SIGMA and INITIAL_SFC parts in the gsm script to specify the files.
Sure, it should not cause peoblem, except that the date record in the forecast SST must be verifying date. If it contains initial date and forecast hour, it needs to be converted using incdte and force_grib_date_mon in libs/etc. You may also need to concatinate all the SST files if they come in separate files. If you have problem in the conversion, post your question again.