Figure 7.1. Ten-year running mean surface temperature anomalies for the Tropical Atlantic Main Development Region and global average. The MDR is defined here as the region 10o N-20o N, 80o W-20o W. The dataset used is the combined land-ocean HadCRUT3. Anomalies (in degrees Celsius) have been adjusted to have zero mean for the period 1881-1920.

Low-frequency surface temperature variability in the MDR is dominated by sea surface temperature variability in this predominantly oceanic region. The surface temperature data indicate that the MDR warmed by several tenths of a degree Celsius over the twentieth century (Figure 7.1). This warming has been noted previously (e.g., Emanuel, 2005a; Trenberth, 2005; K06; Santer et al, 2006). The MDR warming roughly tracks the increase in global mean surface temperature, but with larger multidecadal swings in temperature compared with those in the global mean record. Concerning the global mean temperature increase, a large body of research has assessed the possible role of increasing greenhouse gases on global mean temperature (e.g., Meehl et al, 2004; International Ad Hoc Detection and Attribution Group [IADAG], 2005; K06) and concludes that most of the global warming over the past 50 years is likely due to the increase in greenhouse gases. Fewer studies have addressed this question on regional scales such as the MDR. Santer et al. (2006) recently examined warming trends in both the Atlantic and Pacific tropical cyclogenesis regions using 22 different climate models and concluded that there is an 84% chance that external forcing led to at least two-thirds of the observed SST increases in these regions.

In another recent model-based assessment (K06) observed twentieth-century surface temperature trends in the MDR and various other regions were compared with trends obtained by using two new Geophysical Fluid Dynamics Laboratory (GFDL) global climate models. Three different historical climate-forcing scenarios were examined (see below). Current best estimates of a number of historical climate forcings over the period 1860-2000 were specified for these scenarios. The forcings included representations of greenhouse gases, volcanic eruptions, solar variability, land cover changes, and aerosols. These forcings were incorporated more realistically than those used in previous GFDL coupled climate model experiments (e.g., Broccoli et al., 2003). The aerosol forcing included the ''direct effect'' only and did not include effects of interactions of aerosols with clouds or precipitation processes.

Figure 7.2 shows late summer sea surface temperatures in the MDR as simulated in the K06 historical runs compared with observations from the HadlSST (Rayner et al., 2003) and ERSST.v2 (Smith and Reynolds, 2003, 2004) datasets. The top panel (Figure 7.2a) shows the observed MDR series in comparison to an eight-member ensemble of experiments, which used both anthropogenic and natural historical forcings (i.e., all available forcings). Natural forcings (Figure 7.2b) include only volcanic aerosols and long-term variability of solar radiation. Anthropogenic forcings (Figure 7.2c) include only changes in well-mixed greenhouse gases, ozone, aerosols, and land cover. Further details on these forcings are provided in K06. The differences among the ensemble members for each forcing scenario reflect internal climate variability as simulated by the model. Each ensemble member is initialized with different ocean initial conditions taken from a multi-century, 1860-condition control run, and thus begins in a different phase in terms of internally generated modes of the model, such as the model's El Nino, North Atlantic Oscillation (NAO), or internal Atlantic Ocean variability.

In the MDR, the observed long-term warming during the twentieth century is much more realistically simulated in the model runs, which include anthropogenic forcing (i.e., the all-forcing or anthropogenic-only-forcing scenarios) than in the natural-forcing runs. This is particularly true for the late twentieth-century warming. There is some resemblance of the temporal structure of the all-forcing ensemble mean compared to the observations beyond just the century-scale linear trend. However, the observations still exhibit pronounced multidecadal departures from the ensemble mean of the all-forcing runs. The anthropogenic-forcing ensemble mean response (c; n = 4) in the MDR appears fairly linear. A more nonlinear response (increasing slope over time) is evident for the global mean results (see K06, Fig. 1e). The anthropogenic- and all-forcing runs include only the direct effect of aerosols, and so the total forcing from aerosol changes (including indirect effects) is


Individual ensemble members CM2 Ensemble mean (n=8)


Individual ensemble members CM2 Ensemble mean (n=8)

1860 r

0 0

Post a comment