Nonlinear Functions of Satellite-Measured Spectral Radiance as Estimators of Tropopause Height

This study deals with atmospheric tropopause detection by satellite. It is found that incorrect forecasting of the tropopause height yields a systematic bias in temperature profile estimates which use an algorithm (specifically, the “minimum information” method) operating on satellite-measured spectral infrared radiances and forecast temperature profiles as first guess fields. This bias follows from the general inability of linear satellite-based temperature retrieval schemes to substantially correct “shape” errors in the first guess field, owing to the rather low vertical resolving power of such schemes. A non-linear approach to tropopause detection which makes use of certain selected ratios of measured spectral radiances is shown to have some power for correcting these biases.