The effect of polarization sensitivity of Brewer spectrometers on Direct Sun measurements

Abstract

The polarization sensitivity of a Brewer MKIII double spectrophotometer was measured in the laboratory. We found two major sources of polarization sensitivity. 1) The flat quartz plate as the first optical element alters the polarization state of the transmitted light by Fresnel reflection at oblique incident angles. 2) The internal grating produces almost 100% polarization of the incident light perpendicular to the direction of the ruled grating. The combination of both effects results in a zenith angle (ZA) dependence of the instrument’s sensitivity for unpolarized input such as from Direct Sun measurements. The Brewer is 2% more sensitive at ZA=0° and 10% less sensitive at ZA=80° with respect to normal incidence (ZA=35°). Since the ZA-dependence is independent of wavelength this effect cancels out when calculating wavelength-ratios as used for total ozone retrieval. However the ZA-dependence causes errors when absolute signals at single wavelengths are needed as for aerosol optical depth (AOD) retrievals. Based on our laboratory measurements an overestimation of the Langley extrapolation between 3% and 5% is estimated even at best atmospheric conditions. The ZA-dependence causes 0.025-0.045 overestimation of AOD and an underestimation of the Angstrom exponent. We believe that this effect has not been detected from Brewer AOD-measurements since it is masked by larger uncertainty sources of other nature and AOD-comparisons to other instruments in the short UV-region are rare. Knowing the ZA-dependence it is possible to correct for the ZA-effect. We modified our Brewer by incorporating a depolarizer in its optical path and replacing the flat quartz window by a curved one, so that the input is always at normal incidence, which reduces the ZA-effect.