What Caused the Observed Twentieth-Century Weakening of the Walker Circulation

The Walker circulation (WC) is one of the world's most prominent and important atmospheric systems. The WC weakened during the twentieth century, reaching record low levels in recent decades. This weakening is thought to be partly due to global warming and partly due to internally generated natural variability. There is, however, no consensus in the literature on the relative contribution of external forcing and natural variability to the observed weakening of the WC. This paper examines changes in the strength of theWCusing an index called BoxΔP, which is equal to the difference in mean sea level pressure across the equatorial Pacific. Change in both the observations and in World Climate Research Programme (WCRP) Coupled Model Intercomparison Project phase 3 (CMIP3) climate models are examined. The annual average BoxΔP declines in the observations and in 15 out of 23 models during the twentieth century (results that are significant at or above the 95% level), consistent with earlier work. However, the magnitude of the multimodel ensemble mean (MMEM) 1901-99 trend (-0.10 Pa yr -1) is much smaller than the magnitude of the observed trend (-0.52 Pa yr -1). While a wide range of trends is evident in the models with approximately 90% of the model trends in the range (-0.25 to +0.1 Pa yr -1), even this range is too narrow to encompass the magnitude of the observed trend. Twenty-first-century changes in BoxΔP under the Special Report on Emissions Scenarios (SRES) A1B and A2 are also examined. Negative trends (i.e., weaker WCs) are evident in all seasons. However, the MMEM trends for the A1B and A2 scenarios are smaller in magnitude than the magnitude of the observed trend. Given that external forcing linked to greenhouse gases is much larger in the twenty-firstcentury scenarios than twentieth-century forcing, this, together with the twentieth-century results mentioned above, would seem to suggest that external forcing has not been the primary driver of the observed weakening of the WC. However, 9 of the 23 models are unable to account for the observed change unless the internally generated component of the trend is very large. But indicators of observed variability linked to El Niño- Southern Oscillation (ENSO) and the Interdecadal Pacific Oscillation have modest trends, suggesting that internally variability has been modest. Furthermore, many of the nine 'inconsistent' models tend to have poorer simulations of climatic features linked to ENSO. In addition, the externally forced component of the trend tends to be larger in magnitude and more closely matches the observed trend in the models that are better able to reproduce ENSO-related variability. The 'best' four models, for example, have a MMEM of -0.2 Pa yr -1 (i.e., approximately 40% of the observed change), suggesting a greater role for external forcing in driving the observed trend. These and other considerations outlined below lead the authors to conclude that (i) both external forcing and internally generated variability contributed to the observed weakening of the WC over the twentieth century and (ii) external forcing accounts for approximately 30%-70% of the observed weakening with internally generated climate variability making up the rest. © 2011 American Meteorological Society