Abstract
Radio occultation is a remote sensing method used to obtain atmospheric profiles of density, pressure, and temperature versus altitude. The basic measurement, however, is the change in phase path of a coherent radio signal linking the transmitter and receiver, both located outside the atmosphere, in a limb-sounding geometry. The change in phase as the transmitter and receiver move to new locations, so that the radio signal connecting the two now goes through a different part of the atmosphere, can be used to determine the refractivity versus altitude profile for the atmosphere. The refractivity profile in turn is used to obtain the density, pressure, and temperature profiles, from knowledge of the composition and use of the equation of state and from an assumption of hydrostatic equilibrium.

The traditional formulation of this procedure, which assumes spherical symmetry of the atmosphere for the forward as well as the inverse problems, does not lend itself to a clear understanding of the inversion process. It is for this reason that the resolution kernel (aka point-spread-function) of the procedure had remained undiscovered for the last 30 years, which has resulted in ad-hoc definitions of the resolving power of this technique.

We present a new formulation for the study of radio occultation phase measurements. This formulation yields the resolution kernel directly, thus providing a straightforward tool for the visualization of the inversion process. The resolution kernel clearly demonstrates the manner in which small- scale (but larger than a Fresnel zone) atmospheric structure appears in the retrieved profiles.

In the second part of the presentation we address the issue of large-scale horizontal gradients in the atmosphere which, if unaccounted for, systematically bias the retrieved profiles. We have derived expressions for the resulting bias under general conditions, i.e. no particular trajectory is assumed, and the analysis applies to rare as well as dense atmospheres. Given an estimate of the horizontal gradient (which is available from Global Circulation Models) of refractivity, the resulting bias can be predicted, and hence corrected.