INFLUENCE OF SUBPIXEL-SCALE CLOUD-TOP STRUCTURE ON REFLECTANCES FROM OVERCAST STRATIFORM CLOUD LAYERS

Citation
Ng. Loeb et al., INFLUENCE OF SUBPIXEL-SCALE CLOUD-TOP STRUCTURE ON REFLECTANCES FROM OVERCAST STRATIFORM CLOUD LAYERS, Journal of the atmospheric sciences, 55(18), 1998, pp. 2960-2973
Citations number
24
Categorie Soggetti
Metereology & Atmospheric Sciences
ISSN journal
00224928
Volume
55
Issue
18
Year of publication
1998
Pages
2960 - 2973
Database
ISI
SICI code
0022-4928(1998)55:18<2960:IOSCSO>2.0.ZU;2-Y
Abstract
Recent observational studies have shown that satellite retrieval of cl oud optical depth based on plane-parallel model theory suffer from sys tematic biases that depend on viewing geometry, even when observations are restricted to overcast marine stratus layers, arguably the closes t to plane parallel in nature. At moderate to low sun elevations, the plane-parallel model significantly overestimates the reflectance depen dence on view angle in the forward-scattering direction but shows a si milar dependence in the backscattering direction. Theoretical simulati ons are performed that show that the likely cause for this discrepancy is because the plane-parallel model assumption does not account for s ubpixel-scale variations in cloud-top height (i.e., ''cloud bumps''). Monte Carlo simulations comparing 1D model radiances to radiances from overcast cloud fields with 1) cloud-top height variations but constan t cloud volume extinction, 2) flat tops but horizontal variations in c loud volume extinction, and 3) variations in both cloud-top height and cloud extinction are performed over a approximate to 4 km x 4 km doma in (roughly the size of an individual GAC AVHRR pixel). The comparison s show that when cloud-top height variations are included, departures from 1D theory are remarkably similar (qualitatively) to those obtaine d observationally. In contrast, when clouds are assumed Ant and only c loud extinction is variable, reflectance differences are much smaller and do not show any view-angle dependence. When both cloud-top height and cloud extinction variations are included, however large increases in cloud extinction variability can enhance reflectance differences. T he reason 3D-1D reflectance differences are more sensitive to cloud-to p height variations in the forward-scattering direction (at moderate t o low sun elevations) is because photons leaving the cloud field in th at direction experience fewer scattering events ((low-order scattering ) and are restricted to the topmost portions of the cloud. While refle ctance deviations from 1D theory are much larger for bumpy clouds than for flat clouds with variable cloud extinction, differences in cloud albedo are comparable for these two cases.