Last update: 19 Apr 2005

Notes to self:



NESDIS/IMS snow/ice cover (1/16 bedient grid)

USAF snow depth/ice cover (1/8 bedient grid)

Snow/ice used as NAM initial conditions

1.2.Past week’s images


2.1.About the analysis

This is the daily Northern Hemisphere sea ice coverage/snow depth analysis produced by USAF. The Air Force GRIB file is converted to a binary file at NESDIS/CEMSCS. NCEP reads in the binary file, converts it to its own GRIB file2.

2.1.1.Grid specifications:

Grid size: 512 x 512

Projection: Polar stereographic true at 60 degrees latitude

Resolution: 1/8 bedient3, i.e., 47.625 km at 60 N

Pole Point: (257,257)

Reference longitude: 80 W

2.1.2.GRIB file descriptions

There are two GRIB records in each NCEP file:

  1. Ice coverage (1=yes, 0=no). Parameter 91 (ice concentration) was used for this field, since (at least) in 1998, there was not a GRIB parameter specifically for ice coverage

  2. Snow depth (unit: meters; parameter number: 66)

2.1.3.Valid time

The NCEP GRIB file carries a valid time of 00Z, though the most representative actual valid time is 12Z (i.e.12 hours later than the GRIB PDS indicates. Information from Tom Kopp, by way of Ken M. email of 2001/2/26). This is because the CEMSCS binary file we grib’d from has always had a time stamp of 00Z. The GRIB file currently coming out of the AF might have a later time stamp (I have not seen the AF GRIB file yet).

2.1.4.GRIB conversion time at NCEP

2000 UTC (1500 EST, as of 11 March 2005).

2.1.5.Location of files on CCS4:

NCEP GRIB: /dcom/us007003/$yyyymmdd/wgrbbul/snowdepth.grb

2.2.Scripts/source code that converts the data to GRIB:





2.3.Script that fetched the NESDIS CEMSCS binary file:

The binary USAF snow/ice file is fetched from the NESDIS CEMSCS site by one or more of the following four scripts (not sure which) in /nwprod/ush:






3.1.NESDIS/IMS web site:

3.2.About the resolution

From 1997 to February 2004, NESDIS/IMS produced the snow/ice coverage analysis on the 1/16 bedient (24km) grid. In February 2004, the analysis began to be produced on a 4km grid (dimension: 6144x6144; resolution is exactly 6 times as high as the old 1/16 bedient grid), but NESDIS/IMS continues to map the analysis to the coarser, 24km grid. In the rest of this document, we will be referring to the 1/16 bedient, 24km analysis, because this will be the “terminal analysis” used by the Eta model (the next model might use the finer resolution analysis).

3.3.About the analysis

3.3.1.Grid specifications:

The analysis is on the 1/16 bedient grid, i.e. the grid size is (1024x1024), and it is identical to the 1/8 bedient grid the USAF snow/ice analysis is in, except that the resolution is twice that of the 1/8 bedient grid.

3.3.2.ASCII file from NESDIS/IMS

Each day, usually before 17:30 UTC (22:30 ET), NESDIS/IMS places an ASCII version of the analysis on$yyyy$ddd.asc

where $ddd is the Julian date. This site is accessible via anonymous ftp from mmbsrv and blue/white. In the ASCII file, each grid point is represented by a single-digit value:

0 - south of equator

1 - sea (no ice)

2 - land (no snow)

3 - sea ice

4 - snow

3.3.3.Scripts/source code that converts the ASCII data to GRIB:




3.3.4.GRIB file descriptions

On the NCEP CCS, the GRIB’d IMS snow/ice coverage file is at /dcom/us007003/$yyyymmdd/imssnow.grb. There are two GRIB records in each file:

  1. Ice coverage (1=yes, 0=no). Parameter 91 (ice concentration) was used for this field, since (at least) in 1998, there was not a GRIB parameter specifically for ice coverage

  2. Snow coverage: (100=snow, 0=no snow). Parameter number 238 (percentage snow cover) was borrowed for this field, because (at least) in 1998, there was not a GRIB parameter specifically for binary (yes/no) snow coverage

The file is created at 23UTC (18 EST).



NDAS snow depth/ice coverage fields are updated once a day using the IMS and AFWA snow/ice analyses. The update is done at the beginning of the 06Z cycle of NDAS (i.e. 18Z model time). Between updates, model snow/ice fields evolves, driven by model physics.

The snow/ice analyses are processed and mapped to model grid by a program GRDETA, run prior to the 06Z NDAS.

Calling sequence in GRDETA:





|_SNO16GET (read in IMS snow/ice cover)

|_SNO8GET (read in AF snow depth/sea ice cover)

4.2.NAM/Eta scripts:

/nwprod/scripts/ (as of 4 May 2005).

4.2.1.GRDETA run

In /nwprod/scripts/, the part pertains to GRDETA is

export pgm=nam_grdeta

. prep_step


export XLFUNIT_41=$DATA/snowdepth.grb

export XLFUNIT_42=$FIXnam/nam_rfusaflw

export XLFUNIT_43=$FIXnam/nam_imsmask

export XLFUNIT_44=$DATA/sst14.t12z.nam_grid

export XLFUNIT_45=$DATA/imssnow.grb


export XLFUNIT_99=$DATA/stcice.output

rm fort.*


ln -s -f snowdepth.grb fort.41

ln -s -f imssnow.grb fort.45


$EXECnam/nam_grdeta < $PARMnam/nam_modtop.parm >> $pgmout 2>errfile cycle update

In /nwprod/scripts/, prior to the call to nam_rstupt:

if [ $hh -eq 06 ] ; then

cp $PARMnam/nam_rstupt.parm_newsnow snow.parm


cp $PARMnam/nam_rstupt.parm_cycsnow snow.parm


cp snow.parm $COMOUT/ndas.t${cyc}z.snow.parm

4.3.NGM script

The NGM production script is in /nwprod/scripts/; one script runs both the pre-processing and the forecast.


Around 1997 or 1998: USAF and IMS snow/ice analyses began to be used in Eta and NGM

2002/6/17: AFWA begins producing their snow/ice analysis in GRIB format (Ken M. email, 6/17). NESDIS has put in place software to convert the new USAF GRIB snow/ice analysis into the former non-GRIB format, so as the format change by USAF is transparent to users like us who fetch the analysis from the NESDIS CEMSCS. We have decided to continue using the NESDIS binary version of the AF snow/ice analyses and do our own GRIB'ing to avoid making changes in the models (ETA and NGM).

2002/7: NESDIS/IMS moved their ascii snow/ice analysis from 'duc6' to 'gp12'.

2004/2/23: NESDIS upgrades the IMS snow/ice analysis from 24km to 4km, but will continue to provide the 24km analysis by mapping the 4km analysis to 24km

2005/4/12: At 13Z, AFWA implementes a change to 2 parameters in their grib header for the Air Force SNODEP product (Ken M. email, 2005/4/11). This does not affect processing at NCEP since NESDIS has modified their program to continue producing binary file in the same format, to be placed on CEMSCS


6.1.For AF snow/ice analysis:

Script: /nfsuser/g01/emcsrc/wx22yl/misc.daily/pltafsno

Source code: /nfsuser/g01/wx22yl/ncarplot/pltafsno.f

The script is run at 19:15 ET. It also sends PRD.SPPROD.SNODEPH.NHMAMAP (the NESDIS/CEMSIS file, renamed snowdepth in ftp) and snowdepth.grb to

6.2.For IMS snow/ice analysis:

Script: /nfsuser/g01/emcsrc/wx22yl/misc.daily/pltimssno

Source code: /nfsuser/g01/wx22yl/ncarplot/pltimssno.f

The script is run at 19:30 ET. It sends imssnow.grb to It also sends a copy (renamed imssnow.$yyyymmdd.grb) to, per the request of John Hodkinson of ECMWF (2004/2/6).

6.3.For snow/ice field in NAM/ETA:

Script: /nfsuser/g01/emcsrc/wx22yl/misc.daily/pltegrd12

Source code: /nfsuser/g01/wx22yl/ncarplot/pltsno_egrd12.f

The script is run at 03:00 ET.


7.1.Air Force snow/ice analysis

At AFWA (George Gayno email, 2005/4/27):


7.2.NESDIS/IMS snow/ice analysis

When the IMS analysis is late/missing on, contact (email from ssdnotes, 20 May 2001)

Contact Persons:


  1. The NESDIS/CEMSCS binary file: did AFWA use to generate it before it started generating their GRIB file, or was CEMSCS generating this file from day 1 from some other AFWA format?

  2. AF snow/ice grib processing time? At 10 Mar 2005 it was 1500 EST (2000 UTC) Check during EDT.

  3. IMS snow/ice grib processing time? At 13 Mar 2005 it was 1800 EST (2300 UTC). Check during EDT.

  4. IMS ascii file creation time? At 13 Mar 2005 it was 22:13 EST (17:13 UTC). Check during EDT.

  5. What is considered valid time for the IMS analysis? In the GRIB file we set it to 22:00 UTC. Should it be earlier?

  6. Current IMS snow/ice analysis on 1/64 bedient grid? 1/64 bedient is (4096x4096), but the NIESDIS/IMS site says the resolution is (6144x6144).

1 ETA was officially re-named NAM (North American Mesoscale) at 12 UTC 25 Jan 2005, and EDAS is renamed NDAS. In the future, when the underlying model is changed from ETA to, say, WRF, the names NAM/NDAS will remain the same. In the rest of this document, we shall refer to ETA/EDAS rather than NAM/NDAS, since this document describes how the snow/ice analyses are used in ETA/EDAS. A newer model (future NAM) might well use the snow/ice analyses differently.

2 This round-about way of conversion is for legacy reasons – when we first started using the USAF snow/ice analysis in ETA, around 1998, the analysis was available only in the binary format. Later the AF started producing the analysis in GRIB. NESDIS/CEMSCS converts this GRIB file into the old binary format in support of their programs, and we continue to do our own GRIB conversion from the binary file, rather than making changes to the ETA model to accommodate the AF GRIB file. We anticipate that the replacement model for ETA will be reading in the USAF GRIB file directly.

3 A polar stereographic grid (true at 60 N) with a grid increment of 381 km at 60 N is said to have a resolution of "1 bedient". "1/8 bedient" means the resolution is 47.625 km at 60 N.

4 (NCEP’s) CCS: Central Computer System, currently the IBM white/blue.