Accuracy of the Total Ozone Mapping Spectrometer Algorithm at Polar Latitudes.

It has been noted that for large solar zenith angles (theta_circ > 75^ circ), there is some uncertainty in the retrieval scheme for determining the total column ozone amount using the Total Ozone Mapping Spectrometer (TOMS) instruments. This uncertainty arises because the current look-up table radiances were calculated by a radiative transfer algorithm using an approximate pseudo-spherical atmosphere. The pseudo -spherical code calculates the primary scatter using spherical geometry but higher order scattering is computed for a plane-parallel atmosphere. To test the accuracy of the pseudo-spherical approximation, a new method for numerically solving the equation of radiative transfer in a spherical shell atmosphere including polarization has been developed. This technique uses a Gauss-Seidel iteration scheme to calculate a steady state solution including all significant orders of scattering. For the TOMS instrument which was on Nimbus-7, large solar zenith angles corresponded primarily to high latitudes due to its sun-synchronous noon crossing orbit. Therefore, the accuracy of the algorithm at large solar zenith angles becomes a critical issue particularly for the precise measurement of ozone over polar regions. Comparisons between the pseudo -spherical and the spherical Gauss-Seidel codes show that for solar zenith angles greater than 80^ circ the error introduced by not properly accounting for the sphericity can be significant. Large intensity and ozone differences (5-10%) are possible in the forward (phi = 0^circ) and backscatter (phi = 180^circ ) directions which are caused by the incorrect attenuation of the solar beam. Because of its particular viewing geometry (along phi = 90^circ), the error in the Nimbus-7/TOMS ozone amount is generally less than 1%. However, when theta_ circ = 88^circ with a high surface reflectivity, ozone amounts reported for Nimbus-7/TOMS may be overestimated by up to 8%. The TOMS instrument currently on Meteor-3, because of its less inclined orbit, has a much wider range of viewing geometries. Under extreme conditions, it appears that errors on the order of 10 to 30% may be possible.