36
to represent, with a high degree of fidelity, the interdecadal variability of
surface air temperature over the Northern Hemisphere continents from
that time onward. The gridded monthly surface air temperature data
set described in Jones et al. (1985) and Jones (1988) extends back to
that date. Marine surface observations of sea surface temperature (SST),
surface air temperature, and other meteorological variables, derived from
the log books of merchant ships, are also available in a gridded format,
extending backward in time to about 1870 for SST and wind (Woodruff et
al. 1987, Bottomleyet al. 1993, Parker et al. 1995).
A thin global network of stations with barometers was in place by 1900,
which marks the beginning of a series of subjectively analyzed Northern
Hemisphere daily surface weather maps. The U.S. National Meteorological
Center (NMC) began to produce gridded upper air maps in 1946. Manual
analysis of these charts was supplanted by a primitive form of objective
analysis by the early 60's. The record of visible satellite imagery dates
back to 1967 and more quantitative outgoing longwave radiation measurements began in 1974 and have been continuous since that time, apart from
a break in 1978. The newly available satellite observations, together with
increasingly sophisticated multivariate data assimilation made it possible
to extend operational daily surface and upper air analyses to include the
tropics and the Southern Hemisphere. The year 1979, which corresponds to
the Global Weather Experiment and the first year of operation of the European Centre for Medium-Range Weather Forecasting (ECMWF) marks the
effective start date for reliable global upper air data. Coincidentally, it also
marks the beginning of the global temperature and oceanic precipitation
records derived from the microwave sounding unit (MSU) carried aboard
the TIROS satellites (Spencer and Christy 1990, 1992; Spencer 1993).
A few of the more prominent discontinuities in the records deserve
mention. Relatively few marine surface observations were archived during World Wars I and II. Prior World War II, the dominant method of
inferring sea surface temperature was to measure the temperature of a recently extracted bucket of sea water. Such 'bucket measurements' were
supplanted by direct measurements of thermometers housed in condenser
intakes, located in hot engine rooms. This change resulted in a spurious
upward jump in reported temperatures centered around 1941/42 (Folland
and Parker 1995). In recognition of this problem, nighttime marine air
temperature records are sometimes used to substantiate the existence of
long term trends in SST time series (e.g., Parker and Folland 1991). During
to represent, with a high degree of fidelity, the interdecadal variability of
surface air temperature over the Northern Hemisphere continents from
that time onward. The gridded monthly surface air temperature data
set described in Jones et al. (1985) and Jones (1988) extends back to
that date. Marine surface observations of sea surface temperature (SST),
surface air temperature, and other meteorological variables, derived from
the log books of merchant ships, are also available in a gridded format,
extending backward in time to about 1870 for SST and wind (Woodruff et
al. 1987, Bottomleyet al. 1993, Parker et al. 1995).
A thin global network of stations with barometers was in place by 1900,
which marks the beginning of a series of subjectively analyzed Northern
Hemisphere daily surface weather maps. The U.S. National Meteorological
Center (NMC) began to produce gridded upper air maps in 1946. Manual
analysis of these charts was supplanted by a primitive form of objective
analysis by the early 60's. The record of visible satellite imagery dates
back to 1967 and more quantitative outgoing longwave radiation measurements began in 1974 and have been continuous since that time, apart from
a break in 1978. The newly available satellite observations, together with
increasingly sophisticated multivariate data assimilation made it possible
to extend operational daily surface and upper air analyses to include the
tropics and the Southern Hemisphere. The year 1979, which corresponds to
the Global Weather Experiment and the first year of operation of the European Centre for Medium-Range Weather Forecasting (ECMWF) marks the
effective start date for reliable global upper air data. Coincidentally, it also
marks the beginning of the global temperature and oceanic precipitation
records derived from the microwave sounding unit (MSU) carried aboard
the TIROS satellites (Spencer and Christy 1990, 1992; Spencer 1993).
A few of the more prominent discontinuities in the records deserve
mention. Relatively few marine surface observations were archived during World Wars I and II. Prior World War II, the dominant method of
inferring sea surface temperature was to measure the temperature of a recently extracted bucket of sea water. Such 'bucket measurements' were
supplanted by direct measurements of thermometers housed in condenser
intakes, located in hot engine rooms. This change resulted in a spurious
upward jump in reported temperatures centered around 1941/42 (Folland
and Parker 1995). In recognition of this problem, nighttime marine air
temperature records are sometimes used to substantiate the existence of
long term trends in SST time series (e.g., Parker and Folland 1991). During