Evaluation of the Environmental Protection Agency/National Weather Service Ultraviolet Index Forecast against independent UV measurements: Phoenix Arizona (2000–2006)

The effectiveness of the National Weather Service (NWS) Ultraviolet Index (UVI) forecast is evaluated for one of the largest United States metropolitan areas that annually experiences exceedingly large dosages of ultraviolet (UV) radiation, Phoenix Arizona (AZ). Data were collected from Arizona Desert Testing LLC sited at Wittmann AZ, the Volkswagen Arizona Proving Grounds sited at Maricopa AZ and Atlas Material Testing Technology LLC, DSET Laboratories sited at New River AZ. In general, over 50% of the variance ( r = 0.709) in UV radiation is shared between the datasets from the three independent facilities although the geographic distances between sites are relatively large. Overall correlations between the NWS UVI forecasts and the data from the three independent testing facilities are surprisingly weak when the annual cycle is removed from all datasets with less than 21% of the variance in UV data accounted for by the NWS UVI forecast. The lowest correspondence occurred during the summer months with June displaying the lowest correlations, often insignificant at the 0.05 level. To account for these lower correlations, we considered daily sky cover and air quality observations of particles with a diameter less than 10 µm (PM 10 ). Our analyses suggest that large amounts of PM 10 present during the late spring and early summer are a likely cause for low correspondence between the NWS UVI and UV data in June. Given the unexpectedly low overall correlations, optical depth may need to be better incorporated in the NWS UVI forecast process through pollution forecasts in order to increase the correspondence between the NWS UVI and actual UV irradiance. Copyright © 2010 Royal Meteorological Society