The ability to infer cloud thickness and cloud droplet size from  near-infrared and thermal radiance measurements of the Advanced Very  High Resolution Radiometer has been demonstrated by several authors (e.g. Lin and Coakley, 1993; Baum and Wielicki, 1994; Perez et al., 2000).  We have extended this technique to the Moderate Resolution Imaging Spectroradiometer (MODIS) on board the Terra satellite, using five channels in the wavelength range 3.7 - 12.0 um to retrieve cloud optical thickness, cloud droplet effective radius (reff), mean cloud temperature and the column water vapor path above and below the cloud.

The retrieval algorithm uses a 3 layer atmosphere for the forward radiance calculation.  The top and bottom layers include water vapor absorption calculated using Modtran, while the middle layer contains a plane-parallel homogeneous layer cloud with a Gamma-distributed drop size distribution.  Multiple scattering inside the cloud is calculated using DISORT.

We minimize the difference between satellite-measured and model-predicted brightness temperatures in the five channels using a numerical evolutionary technique called "scatter search" (Glover et al., 2002).  The  search finds the global minimum in the brightness temperature difference vector for the four model parameters (optical thickness, droplet effective radius, mean cloud temperature, and column vapor path above the cloud.

We have performed retrievals on 4 MODIS images for which in-situ aircraft measurements were made during the Dycoms II experiment off the California coast in July, 2000.  Preliminary comparisons of the retrieved and in-situ measured reff distributions show good agreement in the four cases.  In the extended abstract we will report further on these cases and on the sensitivity of the retrieval algorithm to uncertainties in model parameters.