Early work in this area has been reported at the PACS PI meeting (Salt Lake City) and in the published proceedings from the TOGA-95 conference (Melbourne), the Atlantic Climate Change Program PI meeting (Enfield & Mayer, 1995), and at the GOALS Symposium (Atlanta; Enfield and Mayer, 1996). The final results were summarized by Enfield and Mayer (1997) and Enfield (1996). We find that the anomalous sea surface temperature (SSTA) variabilities over large tropical Atlantic regions north and south of the ITCZ are mostly independent of each other at the interannual time scale dominating the 1950-1992 SSTA data (Figure 1). Meridional dipole behavior does exist but is highly seasonal, does not affect basin-wide regions, and does not constitute a large fraction of the interannual Atlantic SST variability.
The connections between Atlantic SSTA and the Pacific ENSO have been greatly clarified (Enfield and Mayer, 1997): (a) the tropical North Atlantic is the region most strongly affected (about 25% of the variability is explained by the Pacific); (b) the SSTA there is forced by modified heat fluxes in the mixed layer, associated with ENSO-related wind speed variations in the NE trades. The process is highly seasonal, focused on the boreal spring season in the tropical North Atlantic, and leads to lagged effects in the ITCZ and tropical South Atlantic during the ensuing summer months (Figure 2). The teleconnected ENSO in the Atlantic has a mainly monopole (symmetric) configuration across the ITCZ, which serves to further obscure what dipole variability exists at interannual frequencies.
Tropical North Atlantic SSTA appears to explain rainfall variability in several regions (Enfield, 1996).: the Caribbean, Central America, northern South America and the southeastern United States (Figure 3). The North Atlantic associations are comparable to those with Pacific SSTA variability over the same region. North Atlantic warmings are associated with rainfall increases over the northern part of South America and decreases over the southeastern US, both opposite to the sense of Pacific warmings; they are also associated with increased rainfall in the eastern Caribbean, for which a Pacific influence is not evident. Because the equatorial Pacific and North Atlantic SSTA variabilities are correlated and affect rainfall in some regions in similar ways (e.g., NE Brazil), there exists the possibility that one ocean or the other is not directly associated with rainfall. In other regions the effects of both oceans are probably direct, because the rainfall associations are of opposite sign while most of the Atlantic SSTA variability is unrelated to the Pacific ENSO.
SST Variability and tropical American Rainfall
Recent work (Enfield and Alfaro, 1998) involves the use of higher resolution and higher quality data sets available for the 1979-1995 period, to study smaller regions of the Americas in greater detail. The data sets include the one-degree, objectively analyzed and merged SSTs produced by NMC (Reynolds and Smith, 1994); the NCEP/NCAR atmospheric reanalysis (Kalnay et al., 1996); and the Xie and Arkin (1996) merged rainfall analyses. Some of the relationships seen in these data sets are best exemplified by singular value decomposition (SVD) analyses between western hemisphere SSTA (130°W-20°E, 50°S-50°N) and rainfall anomalies over the Caribbean and surrounding land regions (130°W-50°W, 0°-40°N). The first mode represents Pacific ENSO variability and its rainfall effects (not shown), consistent with Ropelewski and Halpert (1987) and Enfield (1996), with a modal time series that correlates at more than 0.95 with the NINO3 SSTA index for the equatorial Pacific. The next modes (2, 3, 4) are most clearly related to tropical Atlantic SSTA patterns, with mode two (2) being the strongest and most characteristic (Figure 4). Shown are the homogeneous correlation patterns for the SSTA side (upper panel) and rainfall side (lower panel) of SVD mode 2, plus the vector correlation of the NCEP/NCAR 10-meter wind anomalies with the respective time coefficients (green arrows).
What the modal analysis shows, and what Enfield and Alfaro (1998) conclude, is that the Caribbean (and surrounding region) rainfall response is much stronger with respect to tropical Atlantic SST patterns than with respect to the ENSO pattern. A positive (negative) dipole configuration in the tropical Atlantic SSTA is associated with enhanced (reduced) rainfall and weakened (strengthened) NE trades in the eastern North Pacific and across the lower portion of Central America and Caribbean. Significant spreading of the rainfall response occurs northward into Mexico and northern Central America, and southward into northern South America. These associations are further enhanced if the eastern tropical Pacific (after accounting for ENSO) is mostly cool (warm), suggesting that atmospheric stability over the region is affected by both oceans through basin-basin interactions. The implication is that the reliability of climate forecasts for this region during ENSO events can be improved with knowledge of the tropical Atlantic SSTA state, while forecasts for the out-year (non-ENSO) climate may become feasible. In their analysis of rainy season start and end dates based on daily rainfall records, Enfield and Alfaro (1998) find significant and consistent relationships, with a greater dominance by Atlantic variability and enhancement when the Pacific and North Atlantic are of opposite sign. These results place a high premium on further understanding -- and eventually predicting -- tropical Atlantic SSTA variability.
Global SSTA modes:
Through the above work we conclude that inter-American climate is significantly affected by both ENSO and non-ENSO SSTA variability, by both the Pacific and the Atlantic, and that considerable bridging of the variabilities occurs between the Atlantic and Pacific sectors, linked through tropospheric teleconnections (Enfield and Mayer, 1997; Enfield, 1996; Enfield and Alfaro, 1998). This has led us to do a modal analysis of global SSTA variability using the new Kaplan et al. (1998) global SSTA data set, 1856-1991. Therein we separate what is the global ENSO mode (Figure 5) from what are non-ENSO modes, which contain both interannual and decadal to multidecadal variability unrelated to ENSO (Figure 6a;Figure 6b), all of which display global distributions and inter-basin interactions (Enfield and Mestas, 1998a; Enfield and Mestas, 1998b; Mestas and Enfield, 1998, in preparation). A more detailed summary of Enfield and Mestas (1998a) can be found at the web site for the 1997 NCAR Summer Colloquium on El Niño-Southern Oscillation.
The global mode analysis is relevant to PACS research in several ways. Interannual and decadal-to-multidecadal variabilities project onto the same modes and superimpose on each other between modes. This shows that the use of the ENSO mode as a basis for prediction can probably be substantially improved upon by adding the predictability latent in one or more of the other modes. Additionally, it is dramatically shown in our work that the 1982/83 ENSO event, the largest in the twentieth century prior to 1997, is demoted to third or fourth place when the contributions from Pacific interdecadal and Atlantic multidecadal variabilities are accounted for (Figure 7). Both of those long-time scale modes resulted in persistent warm "background" conditions in the eastern equatorial Pacific that were maximum at the height of the 1983 anomaly and increased its magnitude by as much as 50%.
All of the research we have been doing in several areas has converged on the need to better understand how SSTA variability is linked (through the troposphere) between adjacent ocean basins, which we wish to pursue through the use of tropospheric teleconnection patterns for both ENSO and non-ENSO variability. Secondly, through the use of reanalyzed atmospheric data dating back to the 1950's, we wish to investigate the mechanisms by which these inter-basin SSTA modes affect the climate of the Americas, especially in the tropics. We think that the Atlantic and Pacific SSTA variabilities "handshake" constructively (or destructively) across the Americas to produce more (or less) predictable fluctuations in the tropospheric stability and the easterly wave amplification process, which ultimately affect the tropical rainy season and the many societal activites that depend on it.
Page last modified Feb. 24, 1998
Back to PACS home page