The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Natural Climate Variability on Decade-to-Century Time Scales
(a) Power spectrum of EOF 2 of SST. The thin dashed lines represent the background red-noise spectrum and its 95 percent confidence limit. (b) As in (a) but for EOF I of air temperature.
ture and E1 of SST (not shown) are red, reflecting the dominance of the long trends before and after World War II.
The similarity of the leading EOFs of the independently measured fields of surface air temperature and SST lends credence to the patterns.
We have examined the sensitivity of the leading EOF patterns of air temperature to temporal filtering. In view of the power spectrum of E1 of air temperature, we pre-filtered the data to emphasize (a) the biennial time scale and (b) the decadal time scale, and then computed EOFs. The first EOF of each of the sets of filtered data (not shown) is nearly identical to the first EOF based on unfiltered data. For the biennial time scale, the first EOF accounts for 26 percent of the variance; on the decadal time scale, it accounts for 29 percent. In view of the low-frequency behavior of E2 of air temperature, we computed EOFs based on lowpass filtered data (half-power point at 5 years). The first EOF of the low-pass filtered data (not shown), which resembles the second EOF of the unfiltered data, accounts for 29 percent of the variance of the low-pass filtered data. The correlation coefficient between the time series of E2 (unfiltered) and El (low-pass) is 0.88. Thus, although the first and second EOFs of unfiltered air temperature are not well separated according to the criterion of North et al. (1982)— they explain 21 percent and 17 percent of the variance, respectively—the sensitivity experiments described above suggest that they are distinct modes of variability.
RELATION OF SURFACE TEMPERATURE VARIABILITY TO THE SURFACE ATMOSPHERIC CIRCULATION
How are the dominant modes of variability in surface temperature related to anomalies in the surface atmospheric circulation? Figure 4a shows the patterns of SST and surface wind anomalies regressed on the time series of E2 of SST (a similar picture is obtained if E1 air temperature is used in place of E2 SST). Note that the polarity of E2 SST is opposite from that shown in Figure 1c and that the SST pattern has been smoothed in space with a 3-point binomial filter. The relationship between the surface wind and SST anomalies is primarily local: stronger-than-normal westerlies and trade winds are coincident with cooler-than-normal temperatures, and southerly wind anomalies are associated with warmer-than-normal temperatures. The negative SST anomalies east of Newfoundland are located slightly upstream of the largest wind anomalies where the air-sea temperature differences are largest. The local nature of the wind/temperature relationships suggests that changes in the air-sea fluxes and wind-induced vertical mixing processes contribute to the formation of the SST anomalies. Ocean modeling experiments by Luksch et al. (1990) and Alexander (1990) confirm this interpretation. A somewhat surprising result is that the wind/temperature relationships are similar on decadal and biennial time scales (not shown).
Is the wind pattern shown in Figure 4a a preferred mode of variability in the atmosphere? Figure 4b shows the second combined EOF of sea level pressure and zonal wind, based on winter means for 1900 to 1989. This mode accounts for 15 percent of the variance in the combined fields. The meridional wind pattern was obtained by regressing the meridional wind anomalies on the time series of the com-