Lidar aerosol retrievals from ICESat using a model-based constrained ratio approach

As is well known, aerosol backscatter and extinction profiles cannot be retrieved unambiguously from lidar observations without an assumption linking aerosol extinction and backscatter (e.g., that extinction-to-backscatter ratio S_a is reasonably spatially constant through a solution layer) as well as requiring additional boundary value or parameter specification information (e.g., a layer optical depth or specified value of S_a). For the mid-visible region (~550 nm), S_a typically ranges between ~20 to 80 depending on aerosol type. But simply selecting the mean of S_a values cited in the literature is subject to too much uncertainty (S_a standard deviation ~ 30%) to yield retrievals that are really useful for accurately quantifying aerosol radiative effects. A recent in-depth analysis of the global aerosol solar radiometer network, AERONET, database has defined a relatively few, well defined aerosol types/models that predominately characterize aerosols observed around the world (Cattrall et al., JGR, 110, D10511, 2005). These model aerosols have well bounded S_a standard deviations of ~ 15% or less. However, assuming a specific model/S_a value for a given retrieval, even using climatological/geographic considerations in the model selection, does not assure that the model really applies/that the retrieval is really correct. A Constrained Ratio Aerosol Model-fit (CRAM) approach (Reagan et al., Proc. IGARSS 2004, IEEE, pp. 1940-1943) can be applied to further bound/reduce uncertainty in the retrievals. Specifically, the aerosol models are characterized by spectral ratios (i.e., dual-wavelength, 532 to 1064 nm, ratios of backscatter extinction and S_a), with uncertainty windows, that permit aerosol retrievals to be obtained subject to the constraint that the lidar data yield retrievals with spectral ratio parameters consistent with a given assumed model (or models). From selected dual-wavelength lidar measurements made with the Geoscience Laser Altimeter System (GLAS) on the ICESat mission, profiles of aerosol backscatter and extinction and layer averaged S_a values have been retrieved by successfully applying CRAM to both smoke and dust layers. As these layers were elevated with clean regions below them, this permitted an independent determination of S_a via the self-transmittance retrieval approach, which further substantiated that the layers were well characterized by the assumed dust and smoke models. Other cases, such as Urban/Industrial aerosol types, have also been investigated with the available GLAS data. The paper and presentation will include an overview of the CRAM approach and discussion of the results obtained from these CRAM based retrievals for a number of interesting aerosol cases.