Evaluating Predictability in the Community Earth System Model in Response to the Eruption of Mount Pinatubo

A central goal of climate research is to determine the perceptible effects of climate change on humans; in other words, the regional and decadal scale effects of carbon dioxide forcing. Identifying the most pronounced and long-lasting responses of climate variables to forcing is important for decadal prediction since forcing terms are a source of predictability on those time scales. Powerful volcanic eruptions provide a transient forcing on the climate system, creating a test bed for climate models. In this study, the Mount Pinatubo eruption is simulated in the Community Earth System Model, CESM1.0, for three model configurations: fully coupled T85 (~1^o), land -atmosphere T85, and land-atmosphere T341 (~1/3^o). Ensembles of simulations with and without volcanic aerosols are compared to determine the relative sensitivity and duration of responses. The predictability is quantified using a unitless signal to noise ratio. Results from all three configurations are compared with observations to evaluate model skill. Stratospheric humidity increases predictably, with spatial analysis showing the response is centered over the tropical tropopause. The corresponding geographical pattern of the tropopause temperature increase lends credibility to attribution of the signal to tropopause flux rather than direct injection of water vapor into the stratosphere. The stratospheric water vapor response to the eruption is long-lived in comparison to other signals, providing a source of long-term predictability in the stratosphere. Novel predictable signals are identified, including a decrease in monoterpene and other volatile organic compounds, an increase in land snow depth and extent, and a decrease in lake, soil, and snow heat content. Additionally, in the fully coupled model the standard deviation of the forced ensemble is reduced 8-14% as compared to the unforced ensemble standard deviation. This implies that the model responds to the forcing in similar ways regardless of initial climate state. The Northern Hemisphere winter warming response improves in spatial distribution and strength both at higher resolution and in the fully coupled model. These results motivate the continued drive to higher resolutions and increased model complexity.