Meteorological Data Guide for Geodesists

Jan 21, 1997 (updated Jan 14, 1999)
Katherine J. Quinn
Massachusetts Institute of Technology
katyq@love.mit.edu

Preamble

This is intended to serve as an informal guide to the data sets used in the AGU Fall 96 poster, "GPS Atmospheric Water Vapor Measurements Without the Use of Local Barometers", by K. J. Quinn and T. A. Herring, as well as an ongoing study into atmospheric delay corrections to satellite laser altimeter ranges.
The data sets referenced here have a basic format: there are various meteorological fields (temperature, geopotential height, relative humidity, winds, vorticity, etc) produced on lat/long grids at vertical levels defined by constant pressure surfaces. There are also some surface grids produced, including a surface pressure field. This is not used in our studies, you should be warned that this is generally not a reliable field and is unsuitable for interpolation to GPS sites. The basic method we use to obtain surface pressure is to interpolate upper atmospheric levels of temperature, geopotential height, and relative humidity and then use these interpolated values in a hydrostatic pressure model to obtain surface pressure values. This method is described is an unpublished report on satellite laser altimeter atmospheric delay corrections by K. J. Quinn and T. A. Herring.
The data actually used in the GPS poster were a set of global aviation analyses (explained below) archived by collegues in the MIT meteorology department. The aviation analyses were used instead of the final analyses since that was what was available to us. These MIT archives are not for general use, however I will explain where to obtain the data in real time.
I hope that the following will give you a good start into the world of meteorological data. There is a great deal of data out there on the web as it leads itself to this type of distribution, much like GPS data. This is a distillation of 2 years of searching around on the web for suitable data, may your learning curve be steeper than mine and good luck.

NOAA Anonymous FTP Site - Real Time Data

FTP address: ftp.ncep.noaa.gov

The most important fact about this ftp site is that it is not a data archive. The current observation and analysis/forecast data only remains in place for approximately 24 hours and is then replaced by the next current set. There are no provisions to retrieve old data, if you miss what you want it's tough luck.
That being said, this is a wonderful site. As long as you know where to find what you want since this is also a very confusing site. It contains a wealth of raw observational data from NCEP (National Center for Enviromental Prediction, formally NMC) as well as the forecasts and analyses produced from this data. At this point I'd like to define the difference between "forecasts" and "analyses". They are both the same set of meteorological fields, however analyses are the best fields available for a certain model run and the forecasts are exactly that: forecasts produced foreward in time from each analysis time. If you are doing a historical study, as we did in our GPS poster, you would want to use the analysis fields. If you are trying to do a real time study you may have to use the forecast fields.
The raw observational data are in directories /pub/data.00z and /pub/data.12z, I have not used any of this observational data so you're on your own there. However these directories also contains the 2.5 by 2.5 degree global analyses produced by the NCEP GDAS (Global Data Assimilation System). These are 6 hourly files, named ced1.ganl.t**z.fnl, where ** = 00, 06, 12, 18. This data is also available in another directory at a finer spatial resolution of 1 by 1 degrees (see next paragraph). This coarser grid is produced to match with historical data archived at NCAR (National Center for Atmospheric Research).
The best available global analyses from NCEP are what's called the final analyses. For a brief description of the final analysis run and a listing of all the files in the /pub/fnl directory see the file /pub/fnl/fnl.inventory. As decribed in this file, the 1 by 1 degree grid files are named /pub/fnl/gdas1.T**Z.PGrbFxx, where ** = 00, 06, 12, 18 are the analysis times and xx = 00, 06 are the forcast hours.
Another type of global analysis available from NCEP is the aviation run. This uses exactly the same model as the final analysis run except that it is run at an early time and therefore includes less data. It also produces more forecasts. For a brief description of the aviation run and a listing of all the files in the /pub/avn directory see the file /pub/avn/avn.inventory. As decribed in this file, the 1 by 1 degree grid files are named /pub/avn/avn.T**z/gblav.T**Z.PGrbFxx, where ** = 00, 12 are the analysis times and xx = {00 to 72 by 3 hours steps} are the forcast hours. An informal comparison between the final analyses and the aviation analyses suggest that there are some differences, especially at higher latitudes. The final analyses should be used if possible.
NCEP also produces regional analyses and forecasts for North America at a much finer horizontal resolution. This is called the Mesoscale ETA model. The effective horizontal resolution is 40km and analyses are produced every 3hrs. I haven't used this data yet, for more information see the file /pub/gcp/eta/gcip.doc.
A great online reference to all the NCEP gridded data sets at the NOAA ftp site can be found at http://spsosun.gsfc.nasa.gov/nmc_rev4.html. This goes into more detail about the actual data sources, the models used, the fields available, as well as giving the appropriate references.
Note that all the data files I have referenced at this site are stored in a format called GRIB (GRIdded Binary). This format is widely used in the meteorological community. The next section will describe how to read this it.

GRIB Format

This format is the WMO (World Meteorological Organization) standard for exchanging gridded binary data. If you're interested in the gory detail you can find it in NMC Office Note 388, available in postscript format in the directory /pub/nws/nmc/docs/gribed1 at the NOAA ftp site. The most important section is section 1 (sect1.ps) as it gives tables explaining what the numbers in different fields of the GRIB format correspond to, as well as explaining how the different grids are stored.
The NOAA ftp site also has some codes for reading GRIB format, these can be found under the directory /pub/nws/nmc/codes. However there is a better GRIB reader out there that is much easier to use and highly recommended. It is called wgrib and all the information and codes can be found at http://wesley.wwb.noaa.gov/wgrib.html.

NCAR Data Archive - Historical Data

URL: http://www.scd.ucar.edu/dss/

If you're looking for historical data this is the place. NCAR is the National Center for Atmospheric Research. It is a huge repository of observational data and analyses. It is much better documented and supported than the NOAA ftp site, which essentially isn't. All the information on ordering data (most of it isn't free!) and obtaining codes to read the data is right there and should be fairly self explainitory.
The data set that the geodetic community would be potentially interested in is the NCEP global gridded analyses, information on this can be found at http://www.scd.ucar.edu/dss/datasets/ds082.0.html. These analyses correspond to the 2.5 by 2.5 degree grids of the final analyses mentioned in the section on the NOAA ftp site. Note that they are only archived every 12 hours as opposed to every 6 hours at the NOAA site. A definitive document on atmospheric and oceanographic data sets, especially those available at NCAR, is NCAR Technical Note 404. The html version can be found at http://www.cgd.ucar.edu:80/cas/tn404/, a compressed postscript file is at ftp://ncardata.ucar.edu/docs/data-intro-technote/tn-404.ps.Z.