Centimeter-level accuracy is crucial for Global Positioning System (GPS) baseline measurements to be useful for many geophysical applications. This implies that baseline vector accuracy must be of the order of a few parts in 108 of baseline length for regional geodesy. The latest techniques developed at JPL for analyzing GPS data have indeed resulted in centimeter-level agreement with solutions determined by Very Long Baseline Interferometry (VLBI) in California, for baseline lengths of up to 1000 km.
The techniques we have found most promising for high accuracy geodesy are: (1) carrier phase ambiguity resolution, (2) multi-day orbit determination, (3) stochastic estimation of the zenith tropospheric delay, and (4) simultaneous use of carrier phase and pseudorange. The order of importance depends upon the scale of the network, and the approaches are often synergistic. For example, ambiguity resolution can depend upon the ability of the other techniques to improve precision.
The future of GPS looks bright if one considers that these results have been achieved despite a partial GPS constellation, no global tracking network, and pseudorange data plagued by multipath. A full GPS constellation and a global tracking network will not be realized until the 1990's, but steps are being made in the right direction. A new receiver/antenna prototype at JPL is showing promise of producing pseudorange observables accurate to 5 cm. Two of these receivers participated in the January 1988 CASA UNO experiment, which was managed by JPL in cooperation with about 30 other institutions. The purpose of this experiment was to accurately measure geodetically interesting baselines in South and Central America. Precise orbit determination for this experiment was enabled by tracking the GPS satellites from Australia, New Zealand, Hawaii, American Samoa, North America, and Europe. Results from this experiment should give valuable information on the potential of GPS.