Aerosol Retrievals

The first area of research dealt with aerosols, specifically the impact of aerosols on the limb radiance and polarization with an eye towards how aerosol profiles and properties might be retrieved. This was accomplished through a sensitivity study, the main results of which are summarized below. The information content of three representative wavelength regimes, 340, 475 and 750 nm, was assessed. It was found that due to the decreased Rayleigh scattering, 750 nm has the largest aerosol signal and is best suited to aerosol retrievals. However, shorter wavelengths will contain slightly different information as the larger optical depths ensure that the aerosol signal originates much closer to the instrument. It was also determined that as only two scattering orders were required for convergence at 750 nm, the single- and multiple-scattered radiation fields would be qualitatively similar.

Overall, limb radiance was observed to be sensitive to the aerosol extinction coefficient and optical thickness of the aerosol but was insensitive to the aerosol size distribution and refractive index. This is due to the fact that the radiance is mainly a function of the P₁₁ phase matrix element which varies slowly with aerosol size and refractive index over the ranges relevant to stratospheric aerosols. Two exceptions exist: in the forward and backward scattering directions and when the Rayleigh limit is reached. No attempt was made to vary the shape of the profile, or perturb it at a given height. It was realized that due to the geometry, limited profile information could be extracted.

Polarization was observed to vary with both the extinction coefficient and size distribution, specifically the effective radius and the effective variance, although sensitivity to the latter was quite dependent on the actual value of effective variance. In addition, through the use of single-scattering, polarization maps as a function of effective radius, specific scattering geometries could be identified for which the polarization was either especially sensitive or insensitive. This will enable optimum geometries to be selected in advance. In particular, polarization at the glory, located near scattering angles of 140-170$^{\circ }$, was found to be very sensitive to the effective radius at 750 nm. Polarization varied only marginally with refractive index. By using a more realistic altitude-dependent effective radius, it was determined that the value of retrieved effective radius will be more indicative of the aerosol above the aircraft.

The impact of surface albedo and surface polarization properties were also examined as both are required as input into the model. In order to avoid adding further errors into the retrieval, the albedo should be known to $\pm0.03$ and surface polarization to $\pm0.02$.

Using the results outlined above, an aerosol retrieval algorithm is proposed. As radiance was largely insensitive to size distribution, it can be used to recover the extinction coefficient vertical profile. To increase the vertical range of the retrieved profile, a four-parameter analytic aerosol profile can be used. The four parameters are determined by minimizing the difference between the modelled and measured radiances at 750 nm. Using these results, polarization is then used to determine the effective radius and variance in a similar manner. Based on the estimated uncertainties in the limb radiance and polarization, it appears that the aerosol extinction coefficient profile can be retrieved to an accuracy of 10% and the effective radius to $\pm0.04~\mu$m. The extent to which the effective variance can be retrieved depends on the width of the size distribution.