Journal cover Journal topic
Geoscientific Model Development An interactive open-access journal of the European Geosciences Union
Geosci. Model Dev., 11, 1093-1113, 2018
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.
Development and technical paper
27 Mar 2018
Optimizing UV Index determination from broadband irradiances
Keith A. Tereszchuk1, Yves J. Rochon1, Chris A. McLinden1, and Paul A. Vaillancourt2 1Air Quality Research Division, Environment and Climate Change Canada, Toronto, Ontario, Canada
2Meteorological Research Division, Environment and Climate Change Canada, Dorval, Quebec, Canada
Abstract. A study was undertaken to improve upon the prognosticative capability of Environment and Climate Change Canada's (ECCC) UV Index forecast model. An aspect of that work, and the topic of this communication, was to investigate the use of the four UV broadband surface irradiance fields generated by ECCC's Global Environmental Multiscale (GEM) numerical prediction model to determine the UV Index.

The basis of the investigation involves the creation of a suite of routines which employ high-spectral-resolution radiative transfer code developed to calculate UV Index fields from GEM forecasts. These routines employ a modified version of the Cloud-J v7.4 radiative transfer model, which integrates GEM output to produce high-spectral-resolution surface irradiance fields. The output generated using the high-resolution radiative transfer code served to verify and calibrate GEM broadband surface irradiances under clear-sky conditions and their use in providing the UV Index. A subsequent comparison of irradiances and UV Index under cloudy conditions was also performed.

Linear correlation agreement of surface irradiances from the two models for each of the two higher UV bands covering 310.70–330.0 and 330.03–400.00 nm is typically greater than 95 % for clear-sky conditions with associated root-mean-square relative errors of 6.4 and 4.0 %. However, underestimations of clear-sky GEM irradiances were found on the order of  ∼  30–50 % for the 294.12–310.70 nm band and by a factor of  ∼  30 for the 280.11–294.12 nm band. This underestimation can be significant for UV Index determination but would not impact weather forecasting. Corresponding empirical adjustments were applied to the broadband irradiances now giving a correlation coefficient of unity. From these, a least-squares fitting was derived for the calculation of the UV Index. The resultant differences in UV indices from the high-spectral-resolution irradiances and the resultant GEM broadband irradiances are typically within 0.2–0.3 with a root-mean-square relative error in the scatter of  ∼  6.6 % for clear-sky conditions. Similar results are reproduced under cloudy conditions with light to moderate clouds, with a relative error comparable to the clear-sky counterpart; under strong attenuation due to clouds, a substantial increase in the root-mean-square relative error of up to 35 % is observed due to differing cloud radiative transfer models.

Citation: Tereszchuk, K. A., Rochon, Y. J., McLinden, C. A., and Vaillancourt, P. A.: Optimizing UV Index determination from broadband irradiances, Geosci. Model Dev., 11, 1093-1113,, 2018.
Publications Copernicus
Short summary
To reduce computational costs, ECCC's new method to calculate the UV Index involves scaling and weighting the irradiance contribution of four low-res UV broadbands currently produced by the GEM forecast model. A high-res irradiance spectrum was produced using Cloud-J to create simulated GEM broadbands to calibrate the original GEM broadbands. The scaled GEM broadbands are then weighted accordingly so that the resultant UV Index field emulates the high-res UV Index field calculated from Cloud-J.
To reduce computational costs, ECCC's new method to calculate the UV Index involves scaling and...