Recent studies have suggested that 20% (Sokolik and Toon, 1996) and up to 30 to 50% (Tegen and Fung, 1995) of the total mineral dust in the atmosphere originates from anthropogenic activities, the precise fraction of mineral dust of anthropogenic origin being extremely difficult to determine. Only the radiative forcing from this anthropogenic component is considered as there is no evidence that the naturally occurring component has changed since 1750, although ice core measurements suggest that atmospheric concentrations of dust have varied substantially over longer time-scales (e.g., Petit, 1990). Because mineral dust particles are of a relatively large size and because it becomes lofted to high altitudes in the troposphere, in addition to the short-wave radiative forcing, it may exert a significant long-wave radiative forcing. The global mean short-wave radiative forcing will be negative due to the predominantly scattering nature in the solar spectrum (although partial absorption may lead to a local positive radiative forcing over high surface albedos and clouds) and the global mean long-wave forcing will be positive.

Sokolik and Toon (1996) used a simple box model and neglected forcing in cloudy regions to estimate a short-wave radiative forcing of -0.25 Wm^-2 over ocean and -0.6 Wm^-2 over land, leading to a global forcing of approximately -0.46 Wm^-2. They point out that this is offset to some extent by a positive long-wave forcing. Tegen and Fung (1995) performed a more detailed three-dimensional GCM modelling study of dust aerosol and estimated that approximately 30 to 50% of the total dust burden is due to changes in land use associated with anthropogenic activity. The radiative forcing using this data was estimated by Tegen et al. (1996) to be -0.25 Wm^-2 in the short-wave and +0.34 Wm^-2 in the long-wave, resulting in a net radiative forcing of +0.09 Wm^-2. Updated calculations of the net radiative forcing based on Miller and Tegen (1998) estimate the radiative forcing to be -0.22 Wm^-2 in the short-wave and +0.16 Wm^-2 in the long-wave, resulting in a net radiative forcing of -0.06 Wm^-2. Hansen et al. (1998) perform similar calculations and calculate a net radiative forcing of -0.12 Wm^-2 by assuming a different vertical distribution, different optical parameters and using a different global model. Jacobson (2001) used a multi-component global aerosol model to estimate the direct radiative forcing to be –0.062 Wm^-2 in the short-wave and +0.05 Wm^-2 in the long-wave, resulting in a net radiative forcing of –0.012 Wm^-2. The effects of non-sphericity of the mineral dust are not accounted for in these calculations. Mishchenko et al. (1997) suggest that differences in the optical parameters between model spheroids and actual dust particles do not exceed 10 to 15%, although changes of this magnitude may have a large effect on the radiative forcing (Miller and Tegen, 1998). An example of the geographical distribution of the radiative forcing is shown in Figure 6.7g (data from Tegen et al., 1996) which shows regions of positive and negative forcing. Positive forcing tends to exist over regions of high surface reflectance and negative radiative forcings tend to exist over areas of low surface reflectance. This is due to the dependency of the forcing on surface reflectance and the additional effects of the long-wave radiative forcing.

One problem that needs to be solved is uncertainty in representative refractive indices (Claquin et al., 1998), and how they vary geographically due to different mineral composition of different source regions (e.g., Lindberg et al., 1976). Sokolik et al. (1993) summarise the imaginary part of the refractive index for different geographic regions finding a range (-0.003i to approximately -0.02i) at a wavelength of 0.55 µm, and differences in refractive index in the long-wave from different geographical sources are also reported by Sokolik et al. (1998). Kaufman et al. (2001) use observations from the Landsat satellite coupled with ground-based sun photometer measurements and suggest that Saharan dust has a smaller imaginary refractive index (-0.001i) at 0.55 µm and absorbs less solar radiation than that used in the above modelling studies leading to a much enhanced shortwave radiative forcing. However, the increase is much less in the modelling study of Hansen et al. (1998) who find the net radiative forcing changes from -0.12 Wm^-2 to -0.53 Wm^-2 when dust is treated as conservatively scattering. von Hoyningen-Huene et al. (1999) determine the imaginary part of the refractive index from surface based absorption and scattering measurements and find that a refractive index of -0.005i best fits the observations for Saharan dust, which is in agreement with the values reported by Sokolik et al. (1993). The refractive indices together with the assumed size distributions determine the optical parameters. The radiative forcing is particularly sensitive to the single scattering albedo (Miller and Tegen, 1999). Additional uncertainties lie in modelling the size distributions (Tegen and Lacis, 1996; Claquin et al., 1998) which, together with the refractive indices, determine the optical parameters. Measurements made by the Advanced Very High Resolution Radiometer (AVHRR) by Ackerman and Chung (1992), showed a local short-wave radiative perturbation off the west coast of Africa of -40 to -90 Wm^-2 and a corresponding long-wave perturbation of +5 to +20 Wm^-2 at the top of the atmosphere. Relating instantaneous observational measurements that do not account for the effects of clouds, diurnal averaging of the radiation, the seasonal signal associated with emissions and the fraction of mineral dust that is anthropogenic to the global mean radiative forcing is very difficult. Because the resultant global mean net radiative forcing is a residual obtained by summing the short-wave and the long-wave radiative forcings which are of roughly comparable magnitudes, the uncertainty in the radiative forcing is large and even the sign is in doubt due to the competing nature of the short-wave and long-wave effects. The studies above suggest, on balance, that the shortwave radiative forcing is likely to be of a larger magnitude than the long-wave radiative forcing, which indicates that the net radiative forcing is likely to be negative, although a net positive radiative forcing cannot be ruled out. Therefore a tentative range of -0.6 to +0.4 Wm^-2 is adopted; a best estimate cannot be assigned as yet.

IPCC Homepage