plete solar activity cycles, major ENSO events, land-use changes, and significant increases in greenhouse gases, chlorofluorocarbons, and anthropogenic aerosols. Understanding climate forcings and effects in the last 25 years is a key requirement for securing reliable predictions of future climate change based on forcing scenario studies. Mt. Pinatubo, providing an estimated −3 W m^-2 global mean surface radiative forcing, is a particularly important test case for examining model-based predictions of response to radiative forcing. The observed surface cooling of roughly 0.4°C has been shown to be consistent with model-estimated responses (Hansen et al., 1992, 2002). Model experiments imposing the inferred vertical radiative forcing profile of Mt. Pinatubo have closely reproduced (Kirchner et al., 1999) the expected seasonal pattern of summer continental cooling and winter warming.

Datasets spanning the past 25 years facilitate a comparison of empirical analysis and model simulations of radiative forcings and their effects on climate. Model experiments employing a combination of anthropogenic and volcanic radiative forcing best match the vertical pattern of temperature changes (Santer et al., 2000) and tropopause height changes (Santer et al., 2003b) over the past couple of decades. A multiple regression analysis of the ENSO index (defined by tropical Pacific sea surface temperatures), volcanic aerosols (according to stratospheric optical depth), solar irradiance (from direct space-based observations), and a linear trend has been argued to reproduce a significant fraction of variability in estimated global lower tropospheric temperatures (Douglass and Clader, 2002). In the latter study, the linear trend was attributed to anthropogenic forcing (a combination of greenhouse gas warming and tropospheric aerosol cooling), while a cooling of 0.5°C was inferred for the Pinatubo eruption and 0.1°C cooling for the solar cycle decrease (forcing of 0.2 W m^-2). These latter conclusions must however be treated with caution because other studies using optimal detection approaches indicate that it is difficult to statistically separate the responses to more than two or three distinct natural and anthropogenic forcings even with a century of data (Stott et al., 2001). Moreover, certain indicators used in the study (e.g., ENSO indices) are not physically or statistically independent of the radiative forcings themselves (e.g., Cane et al., 1997; Collins, 2000; Adams et al., 2003; Mann et al., 2005).

Another study of climate change over the past few decades (Hansen et al., 2002) used a general circulation model to estimate forced changes in both surface temperature estimates and the vertical structure of temperature change (the latter as diagnosed from different channels of the MSU satellite observations). The forcings included well-mixed greenhouse gases, stratospheric (volcanic) aerosols, solar irradiance, ozone, stratospheric water vapor, and tropospheric aerosols. The authors found that observed global temperature change during the past 50 years is primarily a response