A regression technique for determining temperature profiles in the upper stratosphere from satellite-measured radiances.

A technique is described whose application makes it possible to almost double the altitude range over which useful temperature profiles may be obtained. For a wide range of atmospheric temperature profiles, the temperature structure between 30 and 55 km may be derived with an expected error of less than 3 C at 30 km to less than 10 C at 55 km, given the temperature structure in the lower 30 km of the profile, and given a consistent and accurate set of radiance observations. For strongly anomalous conditions in the stratosphere, the accuracy of retrievals at tropospheric levels may also be substantially improved.