Most of the paleoclimatic records of climate variability studied as parts of PEP 1 are proxies for either near-surface temperature variations or precipitation/ soil moisture variations over the Americas. The inter-annual and decadal ENSO-like climate variations depicted previously play important roles in determining those variations—roles to be delineated in this section.
Annual-scale temperature, precipitation, and stream-flow variations associated with CT and GR are illustrated in Fig. 5, which shows regression coefficients relating deviations of October-September averages of precipitation and temperatures from normal to CT
(Figs. 5a and 5c) and GR (Figs. 5d and 5f), respectively. Also shown are correlation coefficients between streamflow and CT (Fig. 5b) and GR (Fig. 5e). Correlations are indicated by circles with radii proportional to the coefficients. Sizing each circle with a radius proportional to the correlation (and regression) coefficient makes the area of the circle proportional to the fraction of variance that is associated with the climate index at each site.
Strong spatial similarities between CT- and GR-forced patterns of precipitation are immediately obvious from a comparison of Figs. 5a and 5d. Streamflow variations with CT and GR also are similar (Figs. 5b and 5e) and mostly reflect, and even amplify (Dettinger et al. 2000), basin-scale precipitation variations. Both CT and GR are positively correlated with precipitation and streamflow in the southwestern United States, Central America, Paraguay, Uruguay, and northern Argentina. Both CT and GR are negatively correlated with precipitation and streamflow in much of northwestern North America, in the tropical parts of South America, east of the Andes, and possibly—although data are particularly sparse for this area—in southernmost South America. Composite averages (not shown here) of precipitation and streamflow during negative and positive phases of CT and GR indicate that drier than normal conditions in tropical South America and in the northwestern United States are more extreme during positive phases of both CT and GR and contribute more to the regression relations shown in Fig. 5 than do wet conditions during negative CTs and GRs. Positive values of both CT and GR also contribute more (wetness) to the regression relations for the southwestern United States and Paraguay-Uruguay-Argentina than do negative phases. Negative phases may be more important contributors to the regression relations in the northeastern United States. Dettinger et al. (1998) and Cayan et al. (1998) also found clear similarities between ENSO and decadal precipitation patterns over western North America.
In the midlatitudes of the Northern Hemisphere, the precipitation and streamflow patterns shown in Fig. 5 reflect the equatorward shifts in midlatitude westerlies and storm tracks associated with positive values of CT and GR that were indicated in Fig. 4 and discussed previously. When CT is positive, more storms, precipitation, and streamflow occur in the southern United States and, across the equator, in Paraguay-Uruguay; the northwestern United States and much of western Canada experience less stormy, drier, and warmer conditions.
FIGURE 5 Regression coefficients—B—relating Cool Tongue (CT) index to October-September (a) precipitation, 1904-90, and (c) surface air temperatures, 1904-90; and correlation coefficients—r—between CT and (b) streamflows, with periods of record ranging as long as 1904-90 but commonly on the order of 40 years. Panels (d)-(f) are the same as (a)-(c) except for Global Residual (GR) index. Radii of circles are proportional to the magnitude of regression and correlation coefficients: black for positive relations, white for negative. Circles inset near 30°S, 135°W indicate scale of influences: B is the regression coefficient and r is the correlation coefficient (r = 0.3 is significantly different from zero at the 95% confidence level for the average length of streamflow records).
The patterns of precipitation, streamflow, SST, and circulation anomalies over the Americas associated with CT and GR are remarkably similar. However, there are notable differences in spatial emphasis. Precipitation and streamflow anomalies in the tropical parts of South America and in the subtropical bands across North and South America are larger and more consistent (spatially) in the CT responses than in the GR responses (Figs. 5a and 5b and Figs. 5d and 5e). Precipitation and, especially, streamflow responses to GR are larger in northwestern North America, México, and Central America than are the corresponding CT responses. Again, we are struck by the remarkable inter-hemispheric symmetries of ENSO-like climate variations in the western Americas.
As was expected from their close ties to CT and GR, regression coefficients relating the rotated PC series (tropical PC and North Pacific PC) to precipitation and streamflow (Figs. 6a and 6b and 6d and 6e) are similar to the corresponding patterns for CT and GR. The precipitation and streamflow patterns associated with the two PC series are also remarkably similar to each other, even though the two PCs are uncorrelated time series; the pattern correlation between the precipitation coefficients in Figs. 6a and 6d is +0.41, and the pattern correlation between streamflow correlations in Figs. 6b and 6e is +0.42. Furthermore, three-fourths of the correlations have the same signs in Figs. 6b and 6e. As with CT and GR, positive tropical PC and North Pacific PC excursions are associated with wetter than normal southwestern United States and Paraguay-Uruguay-Patagonia and drier than normal tropical South America, northwestern North America, and southernmost South America. As with CT and GR, the decadal North Pacific PC is more strongly expressed in northwestern North American precipitation and streamflow varia-
ENSO-like Climate Variations on the Americas a
Was this article helpful?