these data are now available to the research community (Jones et al., 1985, 1986a).2 Efforts are under way to extend the climate record back further into the past, by combining the few available direct measurements with a selection of much longer proxy records, as described elsewhere in this volume. These efforts are complemented by statistical studies aimed at determining the minimum spatial sampling requirements for obtaining reliable global and hemispheric means (North and Kim, 1995).
The surface data are supplemented by a continuous record of satellite-based microwave temperature soundings, beginning in 1979, which provide global coverage (Spencer and Christy, 1990, 1992, 1993). Monthly mean gridded data are available for Channels 2 and 4 of the microwave sounding unit (MSU). The former (MSU-2) provide a measure of the mean temperature of the troposphere (surface to 300 hPa) and the latter (MSU-4) the mean temperature of the lower stratosphere centered near the 70 hPa level.
Although spatial patterns have been examined in a few studies (e.g., Jones and Briffa, 1992; Jones and Kelley, 1983; van Loon and Williams, 1976), hemispheric or global mean temperatures have been the focal point for most of the empirical work on interdecadal temperature variations. The emphasis on hemispheric means is justified by the structure of the leading empirical orthogonal function (EOF) of annual mean surface temperature, which is of the same sign throughout most of the Northern Hemisphere (Barnett, 1978). Even more impressive in this respect is the marked similarity between time series of hemispheric means computed separately for land areas, sea surface temperature, and nighttime marine air temperature (Folland et al., 1990) and the time series of Bradley et al. (1987), presented in Figure 1; these series indicate that on the interdecadal time scale, as China goes, so goes the Northern Hemisphere and vice versa. The same is true of other regions of comparable size (see, e.g., Diaz and Bradley, 1995).
The foregoing evidence of spatial consistency notwithstanding, there still remain questions concerning the interpretation of hemispheric temperature trends in surface air temperature. In this brief review, we will consider the following issues:
Are the unseasonably warm wintertime temperatures over the high-latitude continents during the past decade entirely responsible for the observed warming trend in the Northern Hemisphere, or are they a consequence of a persistent pattern of wintertime circulation anomalies, superimposed on a more fundamental warming trend that is affecting the entire hemisphere during all seasons?
Is the zonally averaged temperature signature associated with the El Niño/Southern Oscillation (ENSO) largely a tropical phenomenon, or is it felt globally?
Are the interdecadal trends in surface air temperature and temperatures aloft virtually identical, or is there evidence of systematic trends in the static stability of the troposphere?
The variability of wintertime temperatures over the high-latitude continents (specifically western Canada and Siberia) is substantially larger than anywhere else in the hemisphere, and anomalies in these two regions are influenced by the Pacific-North American (PNA) teleconnection pattern (Gutzler et al., 1988). Wintertime and springtime warming over these regions has been implicated in the recent rise in the mean surface air temperature during the 1980s (Trenberth, 1990; Folland et al., 1990, 1992; Jones and Briffa, 1992). Indeed, the spatial signature of the PNA pattern is clearly evident in the field of 1,000 to 500 hPa thickness anomalies for the 1980s, shown in Figure 2. Comparison of the wintertime mean and the more bland annual mean fields in Figure 2 suggests that this regional pattern is largely a wintertime phenomenon.
Lest the significance of the high-latitude continents be overstated, it should be noted that trends in the PNA pattern and wintertime temperatures over the high-latitude continents do not always determine the sign of the trend in the mean surface air temperature over the Northern Hemisphere. A notable example is the period of the late 1950s and early 1960s, when the mean temperature of the Northern Hemisphere extratropics fell slightly, even though wintertime temperatures over western Canada rose substantially in connection with a reversal in the predominant polarity of the PNA pattern (van Loon and Williams, 1976; Wallace and Zhang, 1993).
Perhaps as a result of regional features such as the PNA