Middle-latitude climates - The Global Climate System: Patterns, Processes, and Teleconnections - page 101

intervals beginning at 0000 UTC. Also the statistical techniques used in the
assimilation model were continually being improved. The past data also
included changes in the observation systems and changes in the number of
stations (or raw observation points). The workshop had two choices: either to
pick a subset of data that remained stable throughout the chosen period, or
to use observational data available at a given time.
The first choice would give the most stable climate and is analogous to the
use of 30-year means that the World Meteorological Organization (WMO)
recommend to describe the climate of a place. The second choice would
provide the most accurate analysis (on the basis that more is better!) for
the period of the reanalysis. The workshop opted for the second choice and so
the National Centers for Environmental Prediction/National Center for
Atmospheric Research (NCEP/NCAR) 40-Year Reanalysis project was
born and would cover the period 1957-96. The data used in NCEP/NCAR-40
consisted of global rawinsonde data obtained from NCEP (1962 onwards),
the US Air Force (1948-70), China (1954-62), the former USSR (1961-78),
Japan and several other countries; COADS (Comprehensive Ocean-
Atmosphere Data Set, Woodruffe et al. 1987 ); aircraft and constant level
balloon data from a variety of sources, mainly post-1962; surface land
synoptic data since 1949; satellite sounder data since 1969; and satellite
cloud drift winds using geostationary meteorological satellites. Once all
these data were assimilated the reanalysis output came in two main formats:
the binary universal format representation (BUFR) of the WMO and the
gridded binary format (GRIB).
The data output is given for 17 standard pressure levels on a 2.58 latitude
2.58 longitude grid at 0000, 0600, 1200 and 1800 UTC each day. In addition,
a variety of monthly means of atmospheric fields are produced. Full details
of NCEP/NCAR-40 are given in Kalnay et al.( 1996 ).
Whilst NCEP/NCAR-40 was being developed it was intended to extend the
reanalysis backward in time to incorporate (the mainly NH) upper air data for
the 1948-57 decade. This hope became a reality in 2001 when the NCEP/
NCAR 50-Year Reanalysis (1948-98) was described by Kistler et al.( 2001 ).
Information and data for both of these reanalyses can be accessed online at
www.cdc.noaa.gov/cdc/reanalysis/reanalysis.shtml and at dss.ucar.edu/pub/
reanalysis/rean_model.html.
Both these reanalyses are used widely by the climatological community in
current research. As an example, out of 118 articles published in the
International Journal of Climatology in 2003, eleven used these NCEP/
NCAR reanalyses. Generally they compared the reanalysis data with other
contemporary data: in the Arctic Ocean with sea level pressure measured by
the Soviet drifting camps (1950-90) (Cullather and Lynch 2003 ); synoptic
weather classifications in Midwestern USA using ''raw'' radiosonde data