Intercomparison of Radiation Transfer Models Representing Direct Shortwave Forcing by Sulfate Aerosols

A study has been conducted, involving 15 models by 12 groups, to compare modeled forcing (change in shortwave radiation budget) due to sulfate aerosol for a wide range of values of particle radius, optical depth, surface albedo, and solar zenith angle (SZA). The models included highand low-spectral resolution models, incorporating a variety of radiative transfer approximations, as well as a line-by-line model. The normalized forcings (forcing per sulfate column burden) obtained with the radiative transfer models were examined and the differences characterized. All models simulate forcings of comparable amplitude and exhibit a similar dependence on input parameters. As expected for a non-light-absorbing aerosol, forcings were negative (cooling influence), except at high surface albedo combined with low SZA. The relative standard deviation of the zenith-angle-average normalized broadband forcing for 15 models was 8% for particle radius near the maximum in magnitude of this forcing (ca. 200 nm) and at low surface albedo. Somewhat greater model-to-model differences were exhibited at specific SZAs. Still greater differences were exhibited at small particle radii, and much greater discrepancies at high surface albedos, at which the forcing changes sign; in these situations, however, the normalized forcing is quite small. Differences among the models arise mainly from differing treatment of the angular scattering phase function, differing wavelength and angular resolution, and differing treatment of multiple scattering. The relatively small spread in these results suggests that the uncertainty in forcing arising from treatment of radiative forcing of a well characterized aerosol at well specified surface albedo is a minor source of uncertainty compared to that from representing other processes influencing direct forcing by anthropogenic sulfate aerosols and anthropogenic aerosols generally. A journal article describing this intercomparison project has recently been accepted for publication (Boucher et al. 1998), and reference should be made to that publication for the details of the study. The project and results are briefly summarized. Introduction Shortwave radiative forcing by direct light scattering by anthropogenic sulfate aerosol has been suggested to be substantial in the context of longwave forcing by anthropogenic greenhouse gases over the industrial period (Charlson et al. 1991, 1992). Relatively simple expressions were used to estimate this forcing based on atmospheric loading (column burden) of sulfate, but such estimates are subject to concern from the perspective of the accuracy of their representation of the radiative transfer. Issues of concern include mass scattering efficiency (m [g sulfate]), upscatter fraction, wavelength dependence, dependence on surface albedo, and linearity in optical depth. Penner et al. (1994) suggested that the uncertainties arising from these issues amounted to a factor of 2. Other contributions to uncertainty in aerosol forcing include loading, geographical distribution, composition, optical properties, and cloud nucleating properties of the aerosol contributing to the “indirect” effect. Uncertainties in aerosol radiative forcing are thought to represent the greatest contribution to uncertainty in climate forcing over the industrial period (IPCC 1996; Schwartz and Andreae 1996). Several groups have recently addressed radiative forcing by sulfate aerosol, but significant unresolved differences remain. Comparison of these studies suggests that part of the difference may be due to differences in the treatment of radiation. However, it is difficult to confirm this because of differences in the approaches. Boucher and Anderson (1995), using a global model, computed aerosol forcing for accumulation-mode sulfate aerosols. Nemesure et al. (1995) reported forcing for column burdens of monodisperse sulfate aerosols. Pilinis et al. (1995) reported forcing for a “global mean” aerosol consisting of fine and coarse modes. Consequently it was necessary to back out the effects of differences in cloud and surface albedo, aerosol size distributions employed, and the like.