Optimum spectral resolution for computing atmospheric heating and photodissociation rates

Abstract

Rapid, reliable and accurate computations of atmospheric heating rates are needed in climate models aimed at predicting the impact of greenhouse gases on the surface temperature. Photolysis rates play a major role in photochemical models used to assess potential changes in atmospheric ozone abundance due to man's release of chlorofluorocarbons. Both rates depend directly on the amount of solar radiation available at any level in the atmosphere. We present a very efficient method of computing these rates in which integration over the solar spectrum is reduced to a minimum number of monochromatic (or pseudogray) problems by appealing to the continuum features of the ozone absorption cross-sections. To explore the resolutions needed to obtain adequate results we have divided the spectral range between 175 and 700 nm into four regions. Within each of these regions we may vary the resolution as we wish. Accurate results are obtained for very coarse spectral resolution provided all cross-sections are averaged by weighting them with the solar flux across any bin. By using this procedure we find that heating rate errors are less than 20% for all altitudes when only four spectral bands across the entire wavelength region from 175 to 700 nm are used to compute the heating rate profile. Similarly, we find that the error in the photodissociation of ozone is less than a few percent when 10 nm resolution is used in the Hartley and Huggins bands (below 330 nm), while an average over the entire wavelength region from 400 to 700 nm yields similar accuracy for the Chappuis band. For integrated u.v. dose estimates a resolution slightly better than 10 nm is required in the u.v.B region (290–315 nm) to yield an accuracy better than 10%, but we may treat the u.v.A region (315–400 nm) as a single band and yet have an accuracy better than 2%.