37
the postwar period, anemometer measurements have gradually supplanted
observations of sea state (interpreted in terms of the Beaufort Scale) for
inference of wind speed. This change in observing practice, together with
the increasing height of the bridge decks of ships over the years, has resulted in a spurious increase in reported wind speeds (Cardone et al., 1990).
Changes in radiosonde instrumentation and/or analysis procedures in the
1950's and early 1960's raise questions concerning the homogeneity of upper air temperature measurements extending back beyond 1963 (Lambert
1990). An extended record of substantially improved upper air analyses is
expected to be available by the end of 1996, as a result of reanalysis projects
currently under-way at NMC and ECMWF (Bengtsson and Shukla 1988).
3 How complex are the spatial patterns associated
with climate variability?
The climate system is widely regarded as having many spatial degrees of
freedom, yet its time dependent behaviour has often been characterized in
terms of, or purported to be inferred from the variations in a single time
series. This apparent paradox reflects the differing attitudes of researchers
with backgrounds in seasonal to interannual climate prediction versus those
with backgrounds in paleoclimate. The former are accustomed to working
with spatially continuous, gridded fields based on data from a sophisticated
observing system. They are inclined to focus on dynamical variables such
as geopotential height, which exhibit richly textured spatial patterns, with
strong compensation between regional anomalies of contrasting polarity. In
contrast, paleo climatologists are accustomed to working with proxy time
series from widely separated sites which are often dominated by similar
features with the same polarity. As the two cultures commingle in the
melting pot of interdecadal variability, the lack of a commonly accepted
paradigm tends to inhibit communication and generate confusion.
Based on dynamical considerations mentioned in section 2.1, it is conceivable that the surface air temperature field could include a substantial
hemispheric or even global mean component that might be evident in suitably smoothed time series from local sites. The hemispheric and global
mean time series published in the IPCC reports suggests (Folland et al.
1990, 1992) exhibit a distinctive pattern of interdecadal variability that is
remarkably well captured in time-filtered data for China alone, as illus-
the postwar period, anemometer measurements have gradually supplanted
observations of sea state (interpreted in terms of the Beaufort Scale) for
inference of wind speed. This change in observing practice, together with
the increasing height of the bridge decks of ships over the years, has resulted in a spurious increase in reported wind speeds (Cardone et al., 1990).
Changes in radiosonde instrumentation and/or analysis procedures in the
1950's and early 1960's raise questions concerning the homogeneity of upper air temperature measurements extending back beyond 1963 (Lambert
1990). An extended record of substantially improved upper air analyses is
expected to be available by the end of 1996, as a result of reanalysis projects
currently under-way at NMC and ECMWF (Bengtsson and Shukla 1988).
3 How complex are the spatial patterns associated
with climate variability?
The climate system is widely regarded as having many spatial degrees of
freedom, yet its time dependent behaviour has often been characterized in
terms of, or purported to be inferred from the variations in a single time
series. This apparent paradox reflects the differing attitudes of researchers
with backgrounds in seasonal to interannual climate prediction versus those
with backgrounds in paleoclimate. The former are accustomed to working
with spatially continuous, gridded fields based on data from a sophisticated
observing system. They are inclined to focus on dynamical variables such
as geopotential height, which exhibit richly textured spatial patterns, with
strong compensation between regional anomalies of contrasting polarity. In
contrast, paleo climatologists are accustomed to working with proxy time
series from widely separated sites which are often dominated by similar
features with the same polarity. As the two cultures commingle in the
melting pot of interdecadal variability, the lack of a commonly accepted
paradigm tends to inhibit communication and generate confusion.
Based on dynamical considerations mentioned in section 2.1, it is conceivable that the surface air temperature field could include a substantial
hemispheric or even global mean component that might be evident in suitably smoothed time series from local sites. The hemispheric and global
mean time series published in the IPCC reports suggests (Folland et al.
1990, 1992) exhibit a distinctive pattern of interdecadal variability that is
remarkably well captured in time-filtered data for China alone, as illus-