CMIP5 Model Assessments of the Ongoing Global Warming Hiatus and Two Extreme 2012 Climate Events

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 5 February 2014: 8:45 AM
Room C102 (The Georgia World Congress Center )
Thomas R. Knutson, NOAA/GFDL, Princeton, NJ; and R. Zhang, A. Wittenberg, and F. Zeng

Several recent climate phenomena are assessed using a 23-model CMIP5 multi-model ensemble. The pause in global warming since 1998 represents an important ongoing test for climate models. We find that the observed global mean temperature trend (1998-2010) is still within the 5th to 95th percentile of the distribution of trends formed by the CMIP5 ensemble. The CMIP5 ensemble is also used to assess two extreme climate events that occurred in 2012. For Arctic sea ice extent, both the trend over 2000-2012 and the 2012 extreme low are outside the 5th to 95th percentile range of the multi-model CMIP5 ensemble that uses historical forcing. These changes are also far outside the corresponding range of the control runs alone, strongly suggesting a role for anthropogenic forcing in the ice loss. Potential causes of the under-prediction of 2012 Arctic ice loss by the models include: underestimation of natural variability in the Arctic; underestimation of the models' Arctic ice response to forcing; errors in or missing historical forcings; or the extreme ice loss may simply have been an extremely rare event. For the eastern U.S., 2012 produced record temperature anomalies in March-May (MAM), according to the HadCRUT4 gridded data set. Based on the CMIP5 model assessment, we find a significant anthropogenic component to a long-term (multi-decadal to century scale) warming in this region during MAM. According to the ensemble mean of the CMIP5 historical runs, and using a baseline period of 1881-1920, we estimate that about one third of the 2012 warm anomaly was attributable to external forcing (principally anthropogenic). Eastern U.S. MAM warm anomalies exceeding those of 2012 occur about 12 times more often in the CMIP5 historical runs than in the control runs, suggesting that external forcing (principally anthropogenic) has substantially increased the risk of occurrence of such an event.