Estimating UV actinic fluxes from shadowband irradiance networks

Quantification of UV radiation in the lower atmosphere is fundamental to understanding the formation of photochemical air pollutants - such as ground-level ozone (O3) and secondary particulate matter (PM) - that cause millions of premature deaths per year and extensive crop damage globally. The chemically relevant radiative quantity is the spectral actinic flux F(λ), photons cm^-2 s^-1 nm^-1 impingent on atmospheric molecules, while the quantity more frequently measured is the spectral irradiance E(λ), W m^-2 nm^-1, as with upward-facing flat radiometers. The USDA/CSU UV-B Monitoring and Research Program has ca. 40 locations across the USA and has been operational since 1993 measuring E(λ) at discrete UV and visible wavelengths. The availability of direct and diffuse components of E(λ) facilitates the estimation of corresponding F(λ). Here, we use a radiative transfer model (TUV) to show this show that, if the sky condition can be identified unambiguously as either clear or fully overcast, simple parameterizations allow estimation of F(λ) from measured E(λ) with an uncertainty of about 10% or better. Intermediate situations, e.g. when the solar beam is substantially but not entirely attenuated, can incur larger uncertainties (up to 20%) and require additional studies. Comparison of such network-estimated UV actinic fluxes to those used in air quality models will be crucial to understand how much of the model uncertainty and bias results from radiative issues, rather than from errors in pollutant emission inventories or other parameterizations.