Surface Albedo Effects

 

Although not a property of the aerosol profile, surface albedo is nevertheless an important input parameter which will influence both radiance and polarization. Until now, an albedo of 0.3 has been used. Comparisons will be made against albedo values of 0.25, 0.28, 0.32, and 0.35. It is expected that the albedo will be known to at least 0.05 from either satellite data or the retrieval process developed using a combination of the nadir and horizontal flux fields (see Section 3.3 and Appendix A). Results are plotted in Figure 5.14.

As long as there is no direct surface component, the changing albedo did not have a large impact on radiances. Comparing with Figures 5.4 and 5.5, a change in albedo of 0.03 is similar to a change in number density of about 1% or a shift in the profile of about 0.2 km. Albedo appears to be more important for polarization as the change between successive curves is about 0.01. While the shape of the polarization does not change with albedo, a change in albedo of 0.03 appear similar to a change in effective radius of <0.01 to 0.03 $\mu$m, depending on the actual value of effective radius (from Figure 5.7). Hence albedo can potentially have a sizeable impact on the retrieved size parameters and should be known to at least $\pm0.03$.

 

Finally, the impact of a depolarized surface is examined. While no surface is, in general, completely depolarizing, it is believed that the polarization of the reflected radiation will be small. The exception would be Fresnel surfaces near Brewster's angle, for example; but these are likely to be rare. The reflected Stokes vector component $Q^{\uparrow}(\tau_1)$ is artificially set to,

$$Q^{\uparrow}(\tau_1)=LP_s \cdot I^{\uparrow}(\tau_1)$$ (9.8)

where LP_s, normally zero, is assigned values of $\pm0.02$ and $\pm0.04$, independent of direction. From the results presented in Figure 5.15, it is clear that there is no impact on radiances. This is expected as the distribution of energy between the two components may change but the total should remain essentially the same. The effects on polarization were larger effect and similar to those observed for albedo. For LP_s=-0.04, the observed polarization increased as the absolute value of $Q(\tau;\mu,\phi)$ (negative for this geometry) was made larger. Similarly, for LP_s=0.04, the observed polarization decreased as this tended to drive $Q(\tau;\mu,\phi)$ closer to zero. The magnitude of the effect is roughly 0.02 between neighbouring curves for downlooking angles and 0.01 between neighbouring curves for uplooking angles. This implies that the polarization of the surface-reflected radiation must be within $\pm0.04$ of zero or errors in the retrieved aerosol quantities may result. It may be possible to use the polarized nadir radiances, along with the radiative transfer model for purposes of atmospheric correction, to determine the polarization of the surface-reflected radiance.

The effects of non-Lambertian surfaces could also be examined. However, as the scan steps which would be most influenced by the type of surface are those with a direct surface component and do not contain much aerosol information, this is not of great importance within the context of this study.