Investigation of the Effects of the Macrophysical and Microphysical Properties of Cirrus Clouds on the Retrieval of Effective Radius

P.W. Stackhouse Jr. and G.L. Stephens
Dept. of Atmos. Sci.
Colorado State University
Fort Collins, CO 80523
303-491-8550
Internet: stack@herschel.atmos.colostate.edu

ABSTRACT




The effects of scattering properties and cloud inhomogeneities on the retrieval of cirrus cloud effective radii (re) are examined using a newly developed Spherical harmonic Spatial Grid Method (SHSG) for two-dimensional radiative transfer. Since bispectral relationships are used to retrieve the re of cirrus cloud particles (eg., Wielicki et al., 1990, Mon. Wea. Rev, 2356 2376) and re has been shown to be an important factor in determining the extent cirrus clouds modulate the climate (Stephens et al., 1990, JAS,1742-1752), it is crucial that the relationships between re and radiative properties be clearly established. The SHSG method developed by Evans (1993, accepted JAS) efficiently computes radiances through clouds with arbitrary inhomogeneities such as those of cirrus clouds and includes an 'independent pixel approximation' which is used to obtain plane-parallel estimations of radiances. A 2D cloud field was obtained from a RHI scan of a 8.66 mm radar for a cirrus cloud sampled on 26 November 1991 during FIRE Cirrus II. The cloud reflectivities were converted to ice water contents (Sassen, 1989, JCAM, 1050-1053) and then scaled to extinction using a measured cirrus size distribution of equivalent area spheres with an inferred ice water content and re = 84 um Simulations were performed at three Landsat wavelengths, 0 83, 1 65, and 2.21 um (Wielicki et al ). Single scattering albedos and asymmetry parameters were made dependent on wavelength only, the latter were selected to correspond with nonspherical ice (Takano and Liou, 1989, 3-19) and used in single and double Henyey-Greenstein phase functions (HG). Therefore, the resulting cloud fields are represented as clouds having a constant re but 2D extinction fields. The solar zenith angle and the solar azimuth angle were set to 60deg. and 180deg. respectively.

The radiances of these cases were normalized to reflectances and plotted bispectrally along with curves of constant re computed using plane-parallel theory from Wielicki et al .. In general, the independent pixel reflectances from SHSG lie on a curve which parallels the theoretical curves of re and roughly correspond to the re = 84 um distribution used for the simulations. However, the double HG function gives reflectances which correspond to re's' as much as 40% larger than 84 um, which results from the fact that the single HG has higher side scattering than the double HG The 2D results indicate that this sensitivity to the phase function decreased with increasing cloud thickness where multiple scattering dominates. The 2D simulations give distributions of radiances which are more scattered and do not follow the curves of constant re in the thinner clouds leads to an parallel theory. In fact, retrieval of re in the thinner clouds leads to an overestimation approaching 40% while for thicker clouds an underestimation of about 25% results where the maximum optical depth approached 4.5. The effect of the cloud geometry for this case is to change the fundamental relationship between the re and the reflectances at the wavelengths shown for this study. These results illustrate that retrieved cirrus re contain significant biases not only due to the lack of understanding of ice scattering properties but also due to the inhomogeneities of the cloud systems themselves.



|Home | FIRE I | FIRE II | FIRE III | FIRE IV CRYSTAL