Correction of Rayleigh scattering effects in cloud optical thickness retrievals

We present results that demonstrate the effects of Rayleigh scattering on the retrieval of cloud optical thickness at a visible wavelength (0.66 mu m). The sensor-measured radiance at a visible wavelength (0.66 mu m) is usually used to infer remotely the cloud optical thickness from aircraft or satellite instruments. For example, we find that without removing Rayleigh scattering effects, errors in the retrieved cloud optical thickness for a thin water cloud layer (tau(c)= 2.0) range from 15 to 60%, depending on solar zenith angle and viewing geometry. For an optically thick cloud (tau(c) greater than or similar to 10), on the other hand, errors can range from 10 to 60% for large solar zenith angles (theta(O) greater than or similar to 60 degrees) because of enhanced Rayleigh scattering. It is therefore particularly important to correct for Rayleigh scattering contributions to the reflected signal from a cloud layer both (1) for the case of thin clouds and (2) for large solar zenith angles and all clouds. On the basis of the single scattering approximation, we propose an iterative method for effectively removing Rayleigh scattering contributions from the measured radiance signal in cloud optical thickness retrievals. The proposed correction algorithm works very well and can easily be incorporated into any cloud retrieval algorithm. The Rayleigh correction method is applicable to cloud at any pressure, providing that the cloud top pressure is known to within +/-100 hPa. With the Rayleigh correction the errors in retrieved cloud optical thickness are usually reduced to within 3%. In cases of both thin cloud layers and thick clouds with large solar zenith angles, the errors are usually reduced by a factor of about 2 to over 10. The Rayleigh correction algorithm has been tested with simulations for realistic cloud optical and microphysical properties with different solar and viewing geometries. We apply the Rayleigh correction algorithm to the cloud optical thickness retrievals from experimental data obtained during the Atlantic Stratocumulus Transition Experiment (ASTEX) conducted near the Azores in June 1992 and compare these results to corresponding retrievals obtained using 0.88 mu m. These results provide an example of the Rayleigh scattering effects on thin clouds and further test the Rayleigh correction scheme. Using a nonabsorbing near-infrared wavelength (0.88 mu m) in retrieving cloud optical thickness is only applicable over oceans, however, since most land surfaces are highly reflective at 0.88 mu m. Hence successful global retrievals of cloud optical thickness should remove Rayleigh scattering effects when using reflectance measurements at 0.66 mu m.