18 messages Bin Guan Fri, Jun 21, 2013 at 2:03 AM To: redacted@redacted Steps to prepare meto data for the recon model: 1. Run get.sh attached to download the data; 2. Prepare the GrADS control file (NLDAS_2012.ctl) according the instructions here: http://www.cpc.ncep.noaa.gov/products/wesley/grib2ctl.html An example .ctl file is attached for your reference. 3. Download save.gs, parsestr.gsf, parseopt.gsf, and qdims.gsf from: http://www.atmos.umd.edu/~bguan/grads/GrADS_Scripts.html 3. Run mkdat.gs attached within GrADS to process the data; the output are binary files that you will copy to the input_meto folder. 3 attachments get.sh mkdat.gs NLDAS_2012.ctl Dominik Schneider Sat, Jun 22, 2013 at 3:25 AM Reply-To: redacted@redacted To: Bin Guan Nice... 1 down, 2 to go. I'm not sure if we have different versions of ipw maybe? but I had to change line 22(?) in get.sh to: for day in `seq -r $startdate,$enddate -d 1` Clarifications: if you compare your NLDAS_2012.ctl and my nldas2012.ctl, the values for xdef and ydef are slightly different....important? Also, dlwrf.ctl the xdef and ydef limits are different than the snodis boundaries i'm used to...this is adjusted in snodis? And what is dlwrf.ctl-e as opposed to dlwrf.ctl? 3 attachments NLDAS_2012.ctl nldas2012.ctl dlwrf.ctl Guan, Bin (3288-Caltech) Sat, Jun 22, 2013 at 6:58 AM To: "" On Jun 21, 2013, at 6:25 PM, Dominik Schneider wrote: Nice... 1 down, 2 to go. I'm not sure if we have different versions of ipw maybe? but I had to change line 22(?) in get.sh to: for day in `seq -r $startdate,$enddate -d 1` Good to know - some shell commands may be platform dependent. Clarifications: if you compare your NLDAS_2012.ctl and my nldas2012.ctl, the values for xdef and ydef are slightly different....important? I can see the difference is due to the number of significant digits. Your xdef/ydef have one less significant digit than mine. I suggest you manually add the additional digit just to be accureate. Also, dlwrf.ctl the xdef and ydef limits are different than the snodis boundaries i'm used to...this is adjusted in snodis? And what is dlwrf.ctl-e as opposed to dlwrf.ctl? No worries, the difference is because in GrADS the Lat/Lon coordinates refer to the center of the pixels, not the boundary. I have no idea what dlwrf.ctl-e is. I never generated such a file. Note that In all these case, the .ctl files are not used anywhere in the model, just produced for our own reference. On Thu, Jun 20, 2013 at 6:03 PM, Bin Guan wrote: Steps to prepare meto data for the recon model: 1. Run get.sh attached to download the data; 2. Prepare the GrADS control file (NLDAS_2012.ctl) according the instructions here: http://www.cpc.ncep.noaa.gov/products/wesley/grib2ctl.html An example .ctl file is attached for your reference. 3. Download save.gs, parsestr.gsf, parseopt.gsf, and qdims.gsf from: http://www.atmos.umd.edu/~bguan/grads/GrADS_Scripts.html 3. Run mkdat.gs attached within GrADS to process the data; the output are binary files that you will copy to the input_meto folder. Dominik Schneider Sat, Mar 29, 2014 at 10:54 PM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" Hi Bin - I hope this email finds you well. I'm back on this to run 2013 but am noticing a difference in my nldas2012.ctl file that I didn't notice before. Specifically, my tdef 8785 linear 23Z31dec2011 1hr zdef 1 linear 1 1 var153sfc 0 153,1,0 ** surface undefined The last looks like your CLWMRsfc 0 153,1,0 ** surface cloud water [kg/kg] - thoughts why this might not have picked up the right name? I could manually change these but want to make sure there isn't smething bigger wrong. The files are attached. Thanks! ds 4 attachments NLDAS_2012.ctl nldas2012.ctl run_grib2ctl.sh grib2ctl.pl Guan, Bin (3288-Caltech) Sat, Mar 29, 2014 at 11:52 PM To: "redacted@redacted" Look for such a line: * produced by grib2ctl v0.9.12.5p45 Is your grib2ctl a different version? Bin Dominik Schneider Sat, Mar 29, 2014 at 11:56 PM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" It is * produced by grib2ctl v0.9.12.6 Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 12:05 AM To: "redacted@redacted" Try the attached version which is what I have been using. 1 attachment grib2ctl.pl Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 12:11 AM To: "redacted@redacted" Also, my wgrib version is v1.8.1.2a (6-11) Bin Dominik Schneider Sun, Mar 30, 2014 at 12:19 AM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" Same result. My wgrib is 1.2.2 Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 12:24 AM To: "redacted@redacted" Looks like a much older version. Might be the reason. Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 12:26 AM To: "redacted@redacted" from grib2ctl.pl: requires wgrib v1.8.0.12 or higher Dominik Schneider Sun, Mar 30, 2014 at 12:27 AM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" seems strange that i have such an old ocpy. how do you check the version number? I was looking at the header Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 12:31 AM To: "redacted@redacted" I also looked at the header. Bin Dominik Schneider Sun, Mar 30, 2014 at 12:36 AM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" we have the same version? iMac:~/Documents/SWEReconstruction/nldas2scripts $ wgrib Portable Grib decoder for NCEP/NCAR Reanalysis etc. it slices, dices v1.8.1.2a (6-11) Wesley Ebisuzaki DWD-tables 2,201-205 (11-28-2005) Helmut P. Frank spectral: Luis Kornblueh (MPI) usage: wgrib [grib file] [options] Inventory/diagnostic-output selections -s/-v short/verbose inventory -V diagnostic output (not inventory) (none) regular inventory Options -PDS/-PDS10 print PDS in hex/decimal -GDS/-GDS10 print GDS in hex/decimal -verf print forecast verification time -ncep_opn/-ncep_rean default T62 NCEP grib table -4yr print year using 4 digits -min print minutes -ncep_ens ensemble info encoded in ncep format Decoding GRIB selection -d [record number|all] decode record number -p [byte position] decode record at byte position -i decode controlled by stdin (inventory list) (none) no decoding Options -text/-ieee/-grib/-bin convert to text/ieee/grib/bin (default) -nh/-h output will have no headers/headers (default) -dwdgrib output dwd headers, grib (do not append) -H output will include PDS and GDS (-bin/-ieee only) -append append to output file -o [file] output file name, 'dump' is default Misc -cmc [file] use NCEP tables for CMC (dangerous) iMac:~/Documents/SWEReconstruction/nldas2scripts $ Guan, Bin (3288-Caltech) Sun, Mar 30, 2014 at 1:06 AM To: "redacted@redacted" I downloaded a few grib files from year 2000 and year 2012 and for them I got the same problem in ctl as yours. I’m not sure if something in these grib files changed as I did not archive my original files. For now I suggest you manually edit the ctl. Bin Dominik Schneider Sun, Mar 30, 2014 at 1:13 AM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" Ok, Thank you for your help. Have a good rest of the weekend. ds Dominik Schneider Tue, Apr 1, 2014 at 1:52 AM Reply-To: redacted@redacted To: "Guan, Bin (3288-Caltech)" Sorry One other thing related to this. gribmap makes an .idx file, can I reuse the old one or should i make a new one despite the strange results from grib2ctl.pl Thanks Bin Guan Tue, Apr 1, 2014 at 2:07 AM To: redacted@redacted What I understand is that the idx file depends on the ctl file. If the latter is wrong the former could potentially be problematic. To make sure you can compare the results based on two idx files, one based on the original ctl, the other based on the manually edited ctl.