Re-evaluation of the ultraviolet hazard action spectrum--the impact of spectral bandwidth.

The action spectra S(lambda) used for risk assessment of ultraviolet (UV) sources has been widely used since its development over three decades ago. This "UV hazard function" is cited in many safety standards, and UV survey instruments are designed to have a fitting spectral response. The UV hazard function was developed to preclude acute injury and minimize chronic effects but was based upon the best available experimental data of that time. With more experimental data and easier computer methods available today, the action spectrum can be re-examined. Therefore, the published experimental thresholds for damage to the cornea and skin were studied as a function of wavelength to revalidate this action spectrum. A key step relates to the proper consideration of the spectral bandwidth used to determine each threshold and the assignment of an effective wavelength for each narrow-band. We methodically analyzed the uncertainties introduced by employing 1-, 5-, and 10-nm spectral bandwidths by comparing published biological threshold data. The errors introduced by using broader bandwidths become enormous within the critical spectral region of 300-320 nm. By convoluting the threshold data and weighting them with trial, or reference, action spectra, it is possible to determine the wavelengths in each bandwidth that are contributing most of the effective dose and correct the experimental action spectrum. It is concluded that the current hazard function S(lambda) remains valid.