This paper describes recent status of the retrievals, and remaining problems. The emphasis is on the statistical analyses of large volumes of satellite data. Satellite retrievals are quality controlled, and examined if physically realistic and self-consistent. Retrievals of t1 and t2 are checked if spectrally coherent. Retrievals of a are tested if reveal a 1/t trend, and increase in scatter, indicative of systematic and random errors, respectively. Probability distribution functions are shown to be log-normal for t1 and t2, and normal for a. Parameters of the respective fits in the VIRS and AVHRR data are compared. Information content of satellite-derived size parameter, a, is analyzed by introducing a signal-to-noise ratio (SNR), h. The SNR is shown to be well approximated as h~t/to. The critical values of to, at which signal in a is comparable to its noise, are compared for the two different sensors.
Supplementary URL: http://orbit-net.nesdis.noaa.gov:80/crad/sat/atm/aerosol/ignatov/index.html