IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis

2.7.2 Explosive Volcanic Activity Radiative Effects of Volcanic Aerosols

Volcanic sulphate aerosols are formed as a result of oxidation of the sulphur gases emitted by explosive volcanic eruptions into the stratosphere. The process of gas-to-particle conversion has an e-folding time of roughly 35 days (Bluth et al., 1992; Read et al., 1993). The e-folding time (by mass) for sedimentation of sulphate aerosols is typically about 12 to 14 months (Lambert et al., 1993; Baran and Foot, 1994; Barnes and Hoffman, 1997; Bluth et al., 1997). Also emitted directly during an eruption are volcanic ash particulates (siliceous material). These are particles usually larger than 2 µm that sediment out of the stratosphere fairly rapidly due to gravity (within three months or so), but could also play a role in the radiative perturbations in the immediate aftermath of an eruption. Stratospheric aerosol data incorporated for climate change simulations tends to be mostly that of the sulphates (Sato et al., 1993; Stenchikov et al., 1998; Ramachandran et al., 2000; Hansen et al., 2002; Tett et al., 2002; Ammann et al., 2003). As noted in the Second Assessment Report (SAR) and the TAR, explosive volcanic events are episodic, but the stratospheric aerosols resulting from them yield substantial transitory perturbations to the radiative energy balance of the planet, with both shortwave and longwave effects sensitive to the microphysical characteristics of the aerosols (e.g., size distribution).

Long-term ground-based and balloon-borne instrumental observations have resulted in an understanding of the optical effects and microphysical evolution of volcanic aerosols (Deshler et al., 2003; Hofmann et al., 2003). Important ground-based observations of aerosol characteristics from pre-satellite era spectral extinction measurements have been analysed by Stothers (2001a,b), but they do not provide global coverage. Global observations of stratospheric aerosol over the last 25 years have been possible owing to a number of satellite platforms, for example, TOMS and TOVS have been used to estimate SO2 loadings from volcanic eruptions (Krueger et al., 2000; Prata et al., 2003). The Stratospheric Aerosol and Gas Experiment (SAGE) and Stratospheric Aerosol Measurement (SAM) projects (e.g., McCormick, 1987) have provided vertically resolved stratospheric aerosol spectral extinction data for over 20 years, the longest such record. This data set has significant gaps in coverage at the time of the El Chichón eruption in 1982 (the second most important in the 20th century after Mt. Pinatubo in 1991) and when the aerosol cloud is dense; these gaps have been partially filled by lidar measurements and field campaigns (e.g., Antuña et al., 2003; Thomason and Peter, 2006).

Volcanic aerosols transported in the atmosphere to polar regions are preserved in the ice sheets, thus recording the history of the Earth’s volcanism for thousands of years (Bigler et al., 2002; Palmer et al., 2002; Mosley-Thompson et al., 2003). However, the atmospheric loadings obtained from ice records suffer from uncertainties due to imprecise knowledge of the latitudinal distribution of the aerosols, depositional noise that can affect the signal for an individual eruption in a single ice core, and poor constraints on aerosol microphysical properties.

The best-documented explosive volcanic event to date, by way of reliable and accurate observations, is the 1991 eruption of Mt. Pinatubo. The growth and decay of aerosols resulting from this eruption have provided a basis for modelling the RF due to explosive volcanoes. There have been no explosive and climatically significant volcanic events since Mt. Pinatubo. As pointed out in Ramaswamy et al. (2001), stratospheric aerosol concentrations are now at the lowest concentrations since the satellite era and global coverage began in about 1980. Altitude-dependent stratospheric optical observations at a few wavelengths, together with columnar optical and physical measurements, have been used to construct the time-dependent global field of stratospheric aerosol size distribution formed in the aftermath of volcanic events. The wavelength-dependent stratospheric aerosol single-scattering characteristics calculated for the solar and longwave spectrum are deployed in climate models to account for the resulting radiative (shortwave plus longwave) perturbations.

Using available satellite- and ground-based observations, Hansen et al. (2002) constructed a volcanic aerosols data set for the 1850 to 1999 period (Sato et al., 1993). This has yielded zonal mean vertically resolved aerosol optical depths for visible wavelengths and column average effective radii. Stenchikov et al. (2006) introduced a slight variation to this data set, employing UARS observations to modify the effective radii relative to Hansen et al. (2002), thus accounting for variations with altitude. Ammann et al. (2003) developed a data set of total aerosol optical depth for the period since 1890 that does not include the Krakatau eruption. The data set is based on empirical estimates of atmospheric loadings, which are then globally distributed using a simplified parametrization of atmospheric transport, and employs a fixed aerosol effective radius (0.42 µm) for calculating optical properties. The above data sets have essentially provided the bases for the volcanic aerosols implemented in virtually all of the models that have performed the 20th-century climate integrations (Stenchikov et al., 2006). Relative to Sato et al. (1993), the Ammann et al. (2003) estimate yields a larger value of the optical depth, by 20 to 30% in the second part of the 20th century, and by 50% for eruptions at the end of 19th and beginning of 20th century, for example, the 1902 Santa Maria eruption (Figure 2.18).

The global mean RF calculated using the Sato et al. (1993) data yields a peak in radiative perturbation of about –3 W m–2 for the strong (rated in terms of emitted SO2) 1860 and 1991 eruptions of Krakatau and Mt. Pinatubo, respectively. The value is reduced to about –2 W m–2 for the relatively less intense El Chichón and Agung eruptions (Hansen et al., 2002). As expected from the arguments above, Ammann’s RF is roughly 20 to 30% larger than Sato’s RF.

Not all features of the aerosols are well quantified, and extending and improving the data sets remains an important area of research. This includes improved estimates of the aerosol size parameters (Bingen et al., 2004), a new approach for calculating aerosol optical characteristics using SAGE and UARS data (Bauman et al., 2003), and intercomparison of data from different satellites and combining them to fill gaps (Randall et al., 2001). While the aerosol characteristics are better constrained for the Mt. Pinatubo eruption, and to some extent for the El Chichón and Agung eruptions, the reliability degrades for aerosols from explosive volcanic events further back in time as there are few, if any, observational constraints on their optical depth and size evolution.


Figure 2.18. Visible (wavelength 0.55 μm) optical depth estimates of stratospheric sulphate aerosols formed in the aftermath of explosive volcanic eruptions that occurred between 1860 and 2000. Results are shown from two different data sets that have been used in recent climate model integrations. Note that the Ammann et al. (2003) data begins in 1890.

The radiative effects due to volcanic aerosols from major eruptions are manifest in the global mean anomaly of reflected solar radiation; this variable affords a good estimate of radiative effects that can actually be tested against observations. However, unlike RF, this variable contains effects due to feedbacks (e.g., changes in cloud distributions) so that it is actually more a signature of the climate response. In the case of the Mt. Pinatubo eruption, with a peak global visible optical depth of about 0.15, simulations yield a large negative perturbation as noted above of about –3 W m–2 (Ramachandran et al., 2000; Hansen et al., 2002) (see also Section 9.2). This modelled estimate of reflected solar radiation compares reasonably with ERBS observations (Minnis et al., 1993). However, the ERBS observations were for a relatively short duration, and the model-observation comparisons are likely affected by differing cloud effects in simulations and measurements. It is interesting to note (Stenchikov et al., 2006) that, in the Mt. Pinatubo case, the Goddard Institute for Space Studies (GISS) models that use the Sato et al. (1993) data yield an even greater solar reflection than the National Center for Atmospheric Research (NCAR) model that uses the larger (Ammann et al., 2003) optical depth estimate.