The Atmospheric Model Intercomparison Project (AMIP) that is supported by the US DOE and run out of the Lawrence Livermore National Laboratory (LLNL) archive climate model output data. At NCAR, there are archives of 5, 10, and 20 year integrations of CCM2. These are kept on their mass storage system. You need an account on their pool of Crays to get to this. There's also daily data from some GCM runs from the early 90's that were archived as part of the MECCA project. Other sites that store model-generated data may be found in the primary document page.
NCEP archives a lot of their observational, weather (gridded output), and reanalysis (gridded output) data at its NOAA-operated data centers, notably the National Climatic Data Center (NCDC) at Asheville NC. There are precipitation datasets, landuse datasets, vegetation databases, et cetera. Also, there are ocean circulation datasets like the Levitus set, and the one Dick Reynolds maintains. Try also the NCAR homepage, and follow the 'data' links. Universities and institutions around the world download and analyze the gridded and observation datasets and put them out for public consumption (see pointers later in the primary document page).
Two people who have played important roles in the development of archive meteorological datasets are Roy Jenne of NCAR, and Jack Woolen of NCEP (jwoolen@noaa.gov). They have been instrumental in, amongst other things, assembling the historical observational data base that is used for reanalyeses; sometimes called "Data Rehabilitation". At the recent First WCRP Workshop on Reanalysis Jenne said that there are 0.5 billion non-satellite weather-related observations collected in the last 50 years (balloons, surface reports et cetera); that's about 30,000 per day on average. The volume is increasing with satellite-retrieved data; now the largest volume of metorologically relevant data comes from the TIROS Operational Vertical Sounder (TOVS) satellites, and more will come online with the Earth Observing System (EOS). If you include TOVS (low density and high density) retrieved data spanning 1979 to 1997, there are an additional 1.28 billion satellite weather-related observations. The total volume therefore is 1.78 billion observations covering the modern era. Although there are inconsistencies in the datasets between NCAR/NCEP, DAO, and ECMWF archives at least volume is not the problem -- you could put the whole archive on a hand full of CDs. People who are dealing with raw observations seem to be less interested in data mining and more interested in the quality of the last 1% of data -- flat files are their friends. At the other end, where reanalyses or climate modelers produce terabytes of high-frequency gridded data there might be more use for data mining and object-oriented data bases. I'm not sure?
Three important assimilated reanalysis data sets reside at:
Other centers around the world are summarized in the primary page of this document. You can view the GSFC Global Change Master Directory.
Go back to Primary Document Page.