The measurement and modeling of broadband UV-A irradiance

Abstract

The estimation of ultraviolet-A radiation across the earth's surface is needed to model plant productivity and future impacts of ultraviolet-B radiation to plant productivity. We have evaluated the quality of broadband ultraviolet-A (UV-A) irradiance measurements within a UV climate monitoring network in the USA and developed a model to estimate the UV-A irradiance from measurements of the global spectral irradiance at 368-nm. The model was developed from ½ hour interval measurements made during 2000 at three locations across the United States and evaluated from ½ hour measurements made during 2000 through 2002 at seven locations. The stability of the UV-A irradiance sensors across the two year period was evaluated by comparison of changes in UV-A sensor response to changes in 368 nm AOD across years on the same (+/-3) day referenced to the change in UV-A response to changes in 368 nm AOD on sequential days during 2000. Most of the seven UV-A sensors installed during 1999 and 2000 appear to have remained stable (within detectable bounds) through 2004. UV-A irradiance was modeled using measured global 368-nm irradiance and empirical functions defining UV-A and 368-nm irradiance relationships derived from a radiative transfer model. The theoretical pseudo two stream discrete ordinates radiative transfer model provided baseline irradiance relationships between UV-A irradiance and 368-nm spectral irradiance. The model estimated the UV-A irradiance at seven locations across the USA with a mean bias error of 0.5 W m^-2 and a root mean squared error of 1.5 W m^-2. The model error was comparable to the combined effect of previously-estimated UV-A and 368-nm irradiance measurement errors but greater than that of the UV-A sensor alone.