Uncertainty analysis of total ozone derived from direct solar irradiance spectra in the presence of unknown spectral deviations

We demonstrate the use of a Monte Carlo model to estimate the uncertainties in total ozone column (TOC) derived from ground-based direct solar spectral irradiance measurements. The model estimates the effects of possible systematic spectral deviations in the solar irradiance spectra on the uncertainties in retrieved TOC. The model is tested with spectral data measured with three different spectroradiometers at an intercomparison campaign of the research project "Traceability for atmospheric total column ozone" at Izaña, Tenerife on 17 September 2016. The TOC values derived at local noon have expanded uncertainties of 1.3% (3.6 DU) for a high-end scanning spectroradiometer, 1.5% (4.4 DU) for a high-end array spectroradiometer, and 4.7% (13.3 DU) for a roughly adopted instrument based on commercially available components and an array spectroradiometer when correlations are taken into account. When neglecting the effects of systematic spectral deviations, the uncertainties reduce by a factor of 3. The TOC results of all devices have good agreement with each other, within the uncertainties, and with the reference values of the order of 282DU during the analysed day, measured with Brewer spectrophotometer #183.