EMC: Mesoscale Eta Model File Info

WD2200 W/NP22

One cycle's worth of forecast output is available in the meso.00z and meso.12z directories. The 00z directory actually holds the 03Z cycle, and the 12z directory holds the 15Z cycle. The files are named...

meso.TXXZ.NAMEFF.tmTT

where XX is the cycle (00,12-see above), NAME is the indicator of the grid and set of variables in the file, FF is the forecast hour (00,03,06...), and TT is the beginning time of the file relative to the start of the forecast. (used mainly for data assimilation periods, tm03 means t=-3h or 00Z for the 03Z cycle)

A list of NAMEs follows along with a brief description of the file contents...

NAME- GRID TYPE (#) - VERT COORD (res)

For users who do not wish to download a dozen 30+ Megabyte files to look at just the 500 mb height from the Mesoscale Eta Model runs, there is an alternative. We are now breaking up the GRIB and BUFR output files from the forecast so that each field is in or station is in its own separate file. This will allow users to transfer only the variables or locations that they are interested in, saving time and file space.

FOR GRIB FILES...

The new files are in separate directories based upon forecast hour and the originating output file. The naming convention is as follows:

/pub/mso/meso.cur/NAME/fxx/PARM_TYPE_LVL_TIME

where:

/pub/mso/meso.cur is the directory containing all of the output from the latest Meso Eta run.

NAME is EGRDSF, EGRD3D, etc., which are subdirectories which contain all of the variables that exist in the meso.TXXZ.NAMEFF.tmTT files.

fxx is a subdirectory defining the forecast hour. The tm3_00 subdirectory contains fields from the initial startup time for the data assimilation phase (tm3 means "t minus 3h") and the tm3_03 subdirectory contains fields from the 3h forecast of the data assimilation phase that began 3h before the start of the actual forecast. The Meso Eta begins at 03 and 15Z, and the data assimilation phase begins at 00 and 12Z, 3h earlier.

PARM is the parameter name and it comes from Table 2, ON388, (GRIB Documentation) the abbreviation column.

TYPE is the name of the type of level (e.g., SFC, ETA, P, Z) and comes from my choice of names using Table 3, ON388 GRIB Documentation. If I couldn't think of a nice name, the number (KPDS(6), 3 digits) is used instead. See Table A below for the list of level type names.

LVL is the value of the level (e.g., 850) which comes straight out of the PDS (KPDS(7)). See Table 3 in the GRIB documentation.

For variables that are valid between certain layers, like storm- relative helicity for example, the top and bottom of the layer is indicated in the file name. LVL will indicate the values of the top and bottom separated by an '='.

For example: RH_SLYR_91=94

indicates the RH between sigma layer .91 and .94, which is one of the Eta's 30mb thick "boundary layers".

TIME is the name of the time range parameter (e.g., AVE, ACC) and again comes from my choice of names using Table 3, ON388 GRIB Documentation. If I couldn't think of a name, the number (KPDS(16), 3 digits) is used instead, See Table 5 GRIB Documentation. If it is an instantaneous value, the _TIME is dropped from the name. See Table B below for the list of time range names.

Some examples:

/pub/mso/meso.cur/EGRDSF/f33/

the 33h forecast directory from the latest run containing the variables on the E-grid near the surface

~/TMP_P_850 850 mb temperature

~/ACPC_SFC_0_ACC Accumulated precipitation

~/RH_AGL_2 2m relative humidity

~/EVP_SFC_0_AVE Surface evaporation, time averaged

~/CSNOW_SFC_0 Categorical rain (precip type)

FOR BUFR FILES... The hourly forecasts from individual stations that are packed into BUFR can be found in:

/pub/mso/meso.cur/BUFR/bufrhr.STATID.YYMMDDHH

Where:

STATID is the 5 digit station id, see the station list to identify non-WMO numbers.

YYMMDDHH is the start date of the forecast.

Reading BUFR

Info on unpacking BUFR files can be found here. Source code for unpacking BUFR files can be obtained from nic.fb4.noaa.gov in /pub/nws/nmc/codes/bufr and /pub/gcp/eta/bufr.read.

Some ideas on ftp'ing and degribbing:

Your ftp scripts are going to have to grow, there's no doubt about it. I suggest setting up similar local file structures to keep the forecast hours separate once you download them. So, you'll be doing alot of cd'ing and lcd'ing in your ftp session.

"mget *" will be much slower than doing separate "get" commands for each specific file.

Once you have a bunch of little files on your local disk, you can cat them together into one big file before degribbing them.

cat /tmp/f33/* > /tmp/bigfile_f33

This bigfile should unpack in the same way as the huge files that are currently available.

TABLE A - Level Type Name

PDS Octet 10 - Name

TABLE B - Time Range Name

PDS Octet 21 - Name