Relating atmospheric aerosol amounts to blue to red ratio and grayscale contrast fluctuations using digitalization of routine webcam photographs taken in the urban environment of Vienna

Visibility and visual contrast depend on several factors such as aerosol concentration, fog attenuation and humidity, viewing direction relative to the sun as well as trace gas amounts. Usually, visibility is determined by observers or by measurement devices such as visiometers. Only few studies reported on the relationship between visibility and aerosol concentration using digital cameras. In the present study, we develop and test new algorithms for the determination of the amount of aerosols by using digital photography. Two years of routine webcam recordings are used for this study. The digital images are taken at the meteorological measurement platform, which is situated on the roof of a university campus building and overlooking the city of Vienna. The photographs are automatically taken as part of Multi AXis Differential Optical Absorption Spectroscopy (MAX-DOAS) measurements, which are continuously performed during daytime at different azimuthal and elevation angles. Two new algorithms for estimating aerosol amounts by quantifying digital images taken on cloud-free days are introduced within the scope of the present study: (1) the ratio of the blue to red (B/R) RGB channels within a vertical pixel line and (2) the grayscale contrast fluctuations (GCF) within a horizontal pixel line. Vertical B/R and horizontal GCF are compared against ground-based aerosol optical depth (AOD) and in-situ particulate matter (PM) < 10 μm (PM10) measurements. Interestingly, the comparisons show promising results for specific azimuthal directions and during certain daytime periods, e.g. for constellations when the sun is to the side or directly behind the camera. In terms of coefficients of determination (R²), B/R correlates best with AOD during summer, while GCF shows better performance in winter. R² values of up to 0.86 and 0.73 are found between horizontal line GCF and in-situ PM10 and between vertical line B/R and AOD, respectively. For a better interpretation of the correlations, simultaneous measurements of relative humidity (RH) are included in the analyses.