Sea Level Pressure Trends: Model-Based Assessment of Detection, Attribution, and Consistency with CMIP5 Historical Simulations

Abstract

AbstractObserved sea level pressure (SLP) trends for 1901–10, 1951–10, and 1981–2010 are assessed using two observed data sources (HadSLP2_lowvar and 20CRv3) compared to a CMIP5 multimodel ensemble. The CMIP5 simulations include runs with (i) no external forcing (Control runs), (ii) natural external forcing only (Natural-Forcing), or (iii) natural plus anthropogenic forcings combined (All-Forcings). We assess whether the CMIP5 All-Forcing ensemble is consistent with observations and whether there is model-based evidence for detectable anthropogenic influence for the observed SLP trends. For the 1901–2010 and 1951–2010 trends, a robustly detectable anthropogenic signal in both observational data products is a zonal band of SLP increase extending over much of the Southern Hemisphere extratropics (30°–50°S). In contrast, the HadSLP2_lowvar and 20CRv3 observed data products disagree on the sign of the century-scale trends in SLP over much of the low-latitude region 25°N–25°S. These differences will limit confident detection/attribution/consistency conclusions for lower-latitude regions, at least until the observational data product discrepancies are better reconciled. The Northern Hemisphere extratropics remains a difficult region for identifying any detectable anthropogenic influence for annual- or seasonal-mean SLP trends. Overall, our results highlight the difficulty in detecting and attributing anthropogenic signals in SLP for relatively short time scales. The observed 1981–2010 regional trends typically have a different pattern and magnitude from the simulated externally forced trends. Consequently, our results suggest that internal variability is likely the dominant driver of most observed 1981–2010 regional trend features, including the pronounced increase in SLP over the central and eastern equatorial Pacific.