AN ALGORITHM FOR INFERRING SURFACE UV IRRADIANCE INCLUDING CLOUD EFFECTS

Recent extratropical ozone depletion and the concomitant increase in s urface ultraviolet (UV) radiation may be expected to adversely influen ce the biosphere. Since few long-term, high quality datasets of surfac e UV are available for assessing these effects, there is a need to dev elop techniques for estimating past levels of biologically harmful UV at a particular location and thus derive long-term trends. This paper presents a semiempirical algorithm, making use of readily available me teorological variables and total column ozone, for inferring historica l UV levels at a particular location, including cloud cover effects. W here input data are available for a network of locations, the techniqu e can be used to generate geographical distributions of surface UV. Me asurements made at Lauder (45.04 degrees S, 169.68 degrees E), from No vember 1993 to October 1994, were used to establish the relationship b etween cloud-induced reductions of erythemal UV and broadband irradian ce, as a function of solar zenith angle-termed cloud cover modifier fu nctions. To demonstrate the performance of the algorithm, these functi ons were used to derive 10-min surface erythemal UV irradiances for ea ch day in November 1994. The algorithm makes use of measured broadband irradiances and clear-sky erythemal and broadband irradiances calcula ted using the following: 1) statistical derivation from measured data and 2) output from a surface spectral irradiance model. Error analysis of the November 1994 surface erythemal UV irradiances indicates an ac curacy of 0.12 +/- 0.69 mu W cm(-2). The percentage error increases wi th solar zenith angle, with an accuracy of 2 +/- 10% for solar zenith angles less than 55 degrees. Although the accuracy deteriorates for la rger solar zenith angles, these errors contribute little to the error in calculated erythemal daily doses except in winter where the daily d oses are small.