Boreal winter surface air temperature responses to large tropical volcanic eruptions in CMIP5 models

We analyzed global surface air temperature (SAT) responses to five major tropical volcanic eruptions from 1870 to 2005 using outputs from 97 historical and 58 Atmospheric Model Intercomparison Project (AMIP) runs that participated in phase 5 of the Coupled Model Intercomparison Project (CMIP5). In observations, there was a 3-yr global cooling trend after the eruption due to reduced shortwave radiation, and a 0.1-K average global-mean SAT recovery, consisting of El Niño–like tropical warming and Eurasian warming, occurred in the first posteruption boreal winter. This global cooling pause was simulated by the multimodel ensemble (MME) mean of the AMIP runs, but not the MME of the historical runs due to the absence of El Niño–like warming. In the historical runs, simulation of El Niño–like warming was influenced by the initial ocean condition (IOC). An El Niño–like response was simulated when the IOC was not in an El Niño state, but the warming was much weaker compared to observations. The Eurasian warming response, despite being reproduced by the MME mean of both AMIP and historical runs, was not as strong as in observations. This is because the simulated positive polar vortex response, an important stratospheric forcing for Eurasian warming, was very weak, which suggests that the CMIP5 models, and even the Climate Forecast System model, underestimate volcanic effects on the stratosphere. Most of the coupled models failed to replicate both the El Niño and the enhanced polar vortex responses, indicating an urgent need for improving air–sea interaction and stratospheric processes in these models.