Decadalmultidecadal Climate Fluctuations

In recent years it has been increasingly recognized that smaller (than ENSO) but significant climate shifts occur on decadal to multidecadal timescales. Such fluctuations may be related to ocean-atmosphere interactions, amenable to simulation and prediction by numerical models. An example in the tropics is the interdecadal succession of comparative rainy periods with drought periods in the Sahel region of sub-Sa-haran Northwest Africa (Nicholson, 1989), along with its strong association with the frequency of Atlantic tropical cyclones (Gray, 1990). Climatic shifts in rainfall and air temperature have also been noted in the Northern Hemisphere extratropics, especially in central and northern Europe and in western North America. These slow climate oscillations correspond to upstream (to the west) fluctuations in the midlatitude westerly wind belts and associated persistent features in atmospheric pressure patterns, such as the Aleutian Low, the Icelandic Low, and the Azores High. Two dominant and quasi-independent oscillations have been noted: the Pacific Decadal Oscillation (PDO) and the NAO. The index time series frequently used to represent the tro-pospheric variability of the PDO and NAO are based on long records of SLP measured near, or interpolated to, the critical nodal points of SLPA patterns.

We are interested in the non-ENSO patterns of variability in SSTA and how they relate to these climate oscillations. Hence, in this section we will show SLPA-based index series for the PDO and NAO with their composite mean SLPA maps over the Northern Hemisphere, and then we will show the corresponding composite maps for SSTA. The composite SSTA maps can then be used as a reference for the non-ENSO global modes discussed in Section 2.5.

For the PDO, the index we have chosen is the DJF av erage of SLPA over the Aleutian Low region, 167.5°-177.5°W, 42.5°-52.5°N. Trenberth and Hurrell (1994) and Latif and Barnett (1996) have used similar Pacific indices. In Fig. 3A, we show the PDO index time series and a smoothed version of it using a loess filter, which is a locally weighted quadratic smoother, with a halfspan of 8 years. Also shown (Fig. 3B) is a polar-projected NH map of the difference between composite means of the winter SLPA data, based on positive and negative values of the index series. Figure 3C shows the K98 SSTA, similarly composited on the PDO index. The corresponding DJF plots for the NAO are shown in Fig. 4. For the NAO index we have chosen to use the difference of normalized SLPA between Lisbon, Portugal, and Stykkisholmur, Iceland, previously used by Hur-rell (1995). Both SLPA indices are signed to be positive

1920 194-0 1960

Years

1920 194-0 1960

Years

FIGURE 3 Composite analysis of the Pacific Decadal Oscillation (PDO) for the December-January-February (DJF) season. (A) The smoothed (dark) and unsmoothed (light) versions of a PDO temporal index defined by averaging sea level pressure anomaly (SLPA) over an area near the Aleutian Islands and 170°W, signed so as to represent the strength of the midlatitude westerly winds in the North Pacific (see text for details). (B) Distribution of the difference between the composite averages of SLPA over the Northern Hemisphere for positive and negative phases of the unsmoothed PDO index. Units are in hectoPascals (hPa). (D) Distribution of the difference of composites of SSTA, similarly calculated for the phases of the PDO index. Units are in degrees Celsius (°C).

FIGURE 3 Composite analysis of the Pacific Decadal Oscillation (PDO) for the December-January-February (DJF) season. (A) The smoothed (dark) and unsmoothed (light) versions of a PDO temporal index defined by averaging sea level pressure anomaly (SLPA) over an area near the Aleutian Islands and 170°W, signed so as to represent the strength of the midlatitude westerly winds in the North Pacific (see text for details). (B) Distribution of the difference between the composite averages of SLPA over the Northern Hemisphere for positive and negative phases of the unsmoothed PDO index. Units are in hectoPascals (hPa). (D) Distribution of the difference of composites of SSTA, similarly calculated for the phases of the PDO index. Units are in degrees Celsius (°C).

when pressures at higher latitudes are negative and the midlatitude westerlies south of the negative pressure node are anomalously strong.

The positive phase of the PDO is characterized by an enhanced Aleutian Low and strengthened westerly winds over the midlatitude central North Pacific (Fig. 3B). There is also some weakening of the polar vortex, but the pattern is neutral over the rest of the Northern Hemisphere. The smoothed index was positive from 1925 -45, negative during 1945 -75, and positive again after the mid-1970s (Fig. 3A). The dominant timescale of the smoothed PDO is shorter (one to three decades) than that of the NAO (>25 years). The high- (low-) index state is associated with greater (less) winter stormi-ness over the central North Pacific and a warmer (cooler) winter climate along the west coast of North America (Latif and Barnett, 1994; Wiles et al., 1998).

The SSTA imprint of the PDO's positive (negative) phase is dominated by an extensive zone of cooler (warmer) SSTs over the Kuroshio Extension region (40°N) east of Japan as far as 145°W (Fig. 3C). This is also consistent with enhanced (diminished) evaporation and mixing under the stronger (weaker) westerly winds, with enhanced (diminished) cooling over that region. An SSTA of opposite sign and equally as large in amplitude is found in the northeastern North Pacific, consistent with observed warmer land temperatures over western North America (Wiles et al., 1998). Accompanying an enhanced Aleutian Low (high PDO index), west coast warmer temperatures are consistent (in an advective sense) with an enhanced poleward ocean circulation in the subarctic gyre of the northeast Pacific and a diminished equatorward California Current System to the south. Corresponding to the North Pacific changes, there is also warming south of the equator in the eastern Pacific, extending poleward to the Chilean coast. Most attempts to explain the SSTA signature of the PDO invoke ocean-atmosphere interactions in the North Pacific, but no clear explanation has yet been proposed for the features off South America.

We note that the overall spatial pattern of SSTAs for the PDO looks very much like the ENSO pattern, but with weaker and broader warming in the eastern tropical Pacific and more intense cooling over the central North Pacific. As with the warm phase of ENSO, the warm eastern Pacific observed since the mid-1970s is associated with a migration of salmon stocks northward from Oregon-Washington to the Gulf of Alaska (Mantua et al., 1997) and a replacement of anchovies by sardines and horse mackerel populations off Peru and Chile (Muck, 1989).

In the high-index state of the NAO (Fig. 4B), the polar vortex is intensified with enhanced low pressures a concentrated in the Greenland region, while pressures are high over the Azores. Westerly winds blow more strongly across the North Atlantic between the two nodes. The NAO has been in its high-index state since about 1970, preceded by low values in the 1950s and 1960s (Fig. 4A). A previous high state during 1905-30 was followed by 25 years of near-normal values. Timescales range from less than a decade to multi-decadal. Under extremes of the NAO, comparatively severe winters (high rainfall) alternately affect northern (high-index) and southern (low-index) Europe (Hur-rell, 1995). The NAO and its climate impacts are consistent with inferences drawn from tree-ring records around the North Atlantic basin (D'Arrigo et al., 1996). In the tropical sector, a high (low) index is associated with less (more) rainfall in the African Sahel (Nicholson, 1989) and less (more) tropical cyclone development west of northwestern Africa (Gray, 1990). That the NAO may now be returning to the low-index phase is suggested by the recent rapid fall in the unfiltered data (Fig. 4A) and is confirmed by a multivariate mode in the updated global SSTA, similar to the Atlantic mode discussed later in Section 2.5 (Landsea et al., 1999).

The winter (DJF) SSTAdistribution corresponding to the high phase of the NAO (post-1970) shows moderately warmer conditions in the western midlatitudes of the North Atlantic (35°-40°N), roughly west of the intensified Azores High and sandwiched between enhanced westerlies to the north and easterlies to the south (Fig. 4C). Cooler temperatures are found under the latter zones of enhanced wind speeds (40°-50°N, 15°-20°N). Although this cooling is consistent with greater evaporation rates at the short timescales, the modeling study of Timmerman et al. (1998) argues that the >35-year timescale is primarily controlled by changes in the thermohaline circulation (THC). At intermediate timescales advection may play a greater role, and for the shorter timescales diabatic heating due to surface fluxes and thermocline entrainment becomes increasingly important (e.g., Bjerknes, 1964; Halliwell, personal communication). The tropical South Atlantic is neutral with respect to the NAO, and the high-latitude South Atlantic is moderately warm when the index is positive. One must be cautious in regard to the last feature, however, because sampling is relatively poor south of 30°S and west of the Greenwich meridian. When considered over all months (not shown), the warming region in the midlatitude North Atlantic is found to be stronger and more zonally extensive, while the warming in the extratropical South Atlantic disappears.

The SSTA signature of the NAO includes an intero-cean feature in the Pacific: when the NAO index is high (low), SSTAs along the U.S. West Coast, in the Gulf of a

FIGURE 4 Same as in Fig. 3, but for a December-January-February (DJF) version of the North Atlantic Oscillation (NAO) index defined by Hurrell (1995).

Alaska, and in the central equatorial Pacific are significantly cooler (warmer). Again, the pattern in the Pacific region resembles the ENSO pattern, but as we point out in Section 2.5, these climate modes do not necessarily separate cleanly, and alternate representations with differing Pacific signatures may be equally, or more, reasonable.

0 0

Post a comment