Temperature dependence of the Brewer ultraviolet data

We characterize the temperature dependence of the U. S. Environmental Protection Agency/University of Georgia network of Brewer spectrophotometers as used to measure solar ultraviolet (UV) radiation. The instruments used in this study are operated in partnership with the National Park Service at 14 national park sites and in 7 urban areas. The daily and seasonal measurements of UV radiation provided by the instruments can be affected by changes in the internal instrument temperatures at the sites. These effects can lead to errors on the order of +/-10% in the resulting spectral data and of the same order of magnitude for CIE-weighted UV. Fortunately, the temperature dependence for each instrument can be quantified and the data corrected, improving the accuracy to values closer to the levels attainable with high-quality calibration and operation. The temperature dependence of the Brewers is found to vary significantly among the different instruments. A 0.8% per degree Celsius dependence can result in temperature effects as large as 12% at sites where temperatures can vary by 15degreesC in 1 day. These effects can result in a +/-5% error in the spectral irradiance. The errors to the spectral irradiance vary seasonally in a manner that is not random: in the warmer summertime the temperature dependence of the instruments can cause the irradiances to be underestimated, while during the colder winters the effect will be to overestimate UV amounts. In the part of the spectrum above 325 nm, the temperature dependence is generally independent of wavelength. Below 325 nm the temperature effects vary as a function of wavelength over a range of values and are generally largest at the shortest wavelengths. Because changes in temperature from one calibration to the next can affect an instrument's response, understanding the temperature effects is necessary to ensure that artificial trends are not introduced into the Brewer data records.