The development of small, high-performance instruments to receive Global Positioning System (GPS) signals has created an opportunity for active remote sounding of the Earth's atmosphere by radio occultation techniques. A prototype demonstration of this capability has been provided by the GPS Meteorology (GPS/MET) experiment, following the first launch of a small Low Earth Orbit (LEO) satellite, MicroLab-1, in April 1995. With known observation geometry specified by the satellites' positions and velocities, the derivative of the phase excess (the Doppler shift excess) characterizes the atmospheric and ionospheric effect on the Doppler frequency shift and can be used to derive the radio occultation bending angle assuming spherical symmetry of the atmospheric refractivity. Ionospheric effects on refraction angles can be eliminated or considerably removed by using two frequencies. The refraction angle profile can then be inverted to obtain the refractivity profile using the Abel transform by invoking the spherical symmetry assumption for the second time. Since atmospheric refractivity depends on pressure, temperature, and specific humidity in a neutral atmosphere, independent knowledge of one of the three quantities allows the other two variables to be retrieved from the GPS refractivity profile. For example, when water vapor contributions to the refractivity are small, a refractivity profile and the hydrostatic equation uniquely define pressure and temperature profiles.

The GPS measurements are not affected by clouds or precipitation and are of high vertical resolution. Instrument calibration is not required and observation errors are statistically independent of the other types of measurements. With these attractive features of GPS measurements, and the promising results of the preliminary GPS-retrieval products, several international projects include plans to launch many more LEO satellites equipped with GPS-receivers. How will this new type of data affect global analyses and possibly change the need for other types of observational data? Before one can answer this question, a method that can extract useful information from these measurements needs to be developed. This talk discusses several physical and computational considerations for the use of GPS occultation observations in numerical weather prediction. We aim at developing a GPS data assimilation method which is accurate, computationally efficient, and feasible for an operational application.