Aerosol climatology using a tunable spectral variability cloud screening of AERONET data

Aerosols, humidity and clouds are often correlated. Therefore, rigorous cloud screening can systematically bias toward less cloudy and drier conditions, underestimating the average aerosol optical thickness (AOT). Here, using AERONET data we show that systematic rejection of variable atmospheric optical conditions can generate such bias in the average AOT. Therefore we recommend two approaches to deal with cloud contamination: (1) to introduce more powerful spectral variability cloud screening and (2) to retain most of the data despite cloud contamination, estimate average cloud contamination and to correct for it. Both methods are applied to aerosol with Angstrom exponent >0.3 and compared with the AERONET cloud screened level 1.5 data. The new methods do not apply for pure dust. Analysis for 10 AERONET stations with similar to 4 years of data, shows almost no change for Rome (Italy), but up to a change in AOT of 0.12 or +30% in Beijing (PRC).