Klobuchar [1991] and Yunck [1993] review the ionospheric effects on GPS positioning and techniques of calibration. While dual frequency correction has been practiced for decades, a few new twists have emerged. Wu and Melbourne [1993] describe a method of combining dual frequency phase and pseudorange to achieve slight noise reduction and a factor of two decrease in data volume when both data types are used. Of chief research interest today are the higher order effects that remain in the dual frequency observables. These can be traced to the interaction between the ionosphere and the earth's magnetic field, and to the differential bending of the L1 and L2 signals by the ionosphere. Brunner and Gu [1991] and Bassiri and Hajj [1993] note that higher order effects can exceed 1 cm under some conditions, with typical values of 1--3 mm. They offer modified dual frequency calibration techniques which may remove as much as 90% of the higher order error. An old technique of single-frequency ionosphere calibration, achieved by combining L1 phase and pseudorange data, was revived by Gold et al [1994] for orbit determination of the Extreme Ultraviolet Explorer. The technique sharply reduced postfit residuals and improved orbit consistency, yielding altitude accuracies of 30--40 cm for a satellite carrying a single-frequency receiver at an altitude of 500 km.
The data bonanza from the GPS global network is stimulating new advances in ionospheric mapping. Until recently, GPS ion mapping has relied on local observations of total electron content (TEC) from individual receivers as they sweep out a band of the ionosphere each day. Mannucci et al [1993] introduced a globally simultaneous technique that features a gridded TEC model with stochastic local TEC adjustment and can produce images of the evolving global ionosphere with arbitrary time resolution. Such techniques can provide precise ionospheric corrections in near real time for single-frequency GPS users. Space based measurements will improve the fidelity and resolution of ionospheric images. Hajj et al [1994] examined through simulation and singular value decomposition the feasibility of 2-D and 3-D tomographic imaging by combining ground and flight GPS data. They found that measurements from space are better suited to both horizontal and vertical ionospheric imaging.
Ionospheric mapping depends on absolute one-way measurements rather than differenced data, imposing a requirement to calibrate the relative L1/L2 instrumental delay biases that would appear as biases in estimated TEC. This is typically accomplished by first calibrating each L1/L2 receiver bias, then solving for the individual satellite biases while estimating TEC [ Coco et al, 1991; Gaposchkin and Coster, 1993]. The global estimation technique permits the solution for individual receiver biases as well. Wilson and Mannucci [1993] report bias estimates with a day-to-day consistency of 0.2--0.4 ns, or about 3 times the precision of previously demonstrated techniques.