The Transport of Volcanic Ash Following an Eruption in a GCM
Kravitz, B., Robock, A., Bourassa, A., Deshler, T., Wu, D., Mattis, I., Finger, F., Hoffmann, A., Ritter, C., Bitar, L., Duck, T.J. and Barnes, J.E.. 2011. Simulation and observations of stratospheric aerosols from the 2009 Sarychev volcanic eruption. Journal of Geophysical Research - Atmospheres 116: D18211, doi:10.1029/2010JD015501.
Volcanism is considered an "external" forcing to the Earth's atmosphere, and, as a forcing process its occurrence is considered to be unpredictable and irregular. However, once the particulate matter and aerosols are injected into the atmosphere it is possible to project the spread of the material using a general circulation model (GCM). In June, 2009, the Sarychev volcano in the Kamchatka Peninsula erupted explosively for a period of approximately five days. At the time, it was the second such eruption within the span of a year. It injected 1.2 Terragrams (Tg) of material into the atmosphere to an estimated height of as much as 16 km.
Kravitz et al. (2011) studied measurements of the optical depth of the aerosol sulfates, and compared these with the projected output from a 20 member ensemble using a GCM, and the goal was to provide suggestions for improving the model's capabilities. The model used was the National Aeronautics and Space Administration (NASA) Goddard Institute for Space Studies Model-E. This is a coupled atmosphere - ocean GCM with fairly coarse resolution in the horizontal (4° by 5° lat/lon) and vertical (23 layers). However, the model contained levels up to 80 km, necessarily including the stratosphere.
The control model run consisted of a 20 member suite globally from 2007 - 2010. In the experiment, 1.5 Tg of volcanic material was injected into the atmosphere at a point near Sarychev in 2008 of the model year. The observed aerosol measurements came from ground based LIDARS at six locations around the world as well as satellite based measurements that profile the aerosol concentration using scattered sunlight (Optical Spectrograph and Infrared Imaging System (OSIRIS)).
The authors found that the model did a reasonably good job in spreading the volcanic material around the Northern Hemisphere, but there were some important discrepancies between the model and observations (e.g. Fig.1). The model transported the material too quickly into the tropics, but too slowly into the high latitudes. The authors speculate that this error may be a function of the need to improve the model stratospheric circulation. Also, the model tended to remove aerosols too quickly from the atmosphere (Fig. 1), especially in the high latitudes, which may have been a function of model overestimate of particulate size. Note that in Fig. 1, the modeled peak aerosol values occur one month earlier than observed and then decrease in concentration too quickly.
Figure 1. Adapted from Fig. 11 from Kravitz et al. (2011). The LIDAR retrievals from Hefei, China as compared to ModelE output and OSIRIS retrievals. (left) The monthly averages of backscatter as a function of altitude maximizing in September 2009. (right) The integrated (15-25 km) optical depth through the stratosphere for the LIDAR data (black), zonally averaged stratospheric aerosol optical depth calculated by the model in the grid latitude containing the Hefei LIDAR (28°-32°N) (red), and OSIRIS retrievals zonally averaged over the latitude band 30°-35°N (blue).
Several articles in the past have highlighted shortcomings of the models used to simulate climate. This is yet another example demonstrating that climate models have a difficult time representing effectively the impact of external-type and seemingly "random" forcing processes which can occur in real time. The likely impact on surface temperature from the Kravitz et al. experiment would be a bias toward warm temperatures on the time scale of months. Thus, it is not likely we can soon improve upon the ability of models to take into account forcing such as volcanic processes in even short-term climate variability without statistically derived parameterizations. Then in the long-term, the best we can do is; "what if"?