The residual tropospheric propagation delay : How bad can it get ? *

The residual range delay error due to an incompletely modelled tropospheric propagation delay can usually be ignored by the average user of wide-area differential GPS. In general the atmosphere is “well behaved” and the use of standard range delay algorithms supplied with standard atmosphere parameter values will adequately model the “average” atmospheric conditions. It is possible however, for the atmosphere to exhibit unusual conditions when the vertical profiles of total pressure or water vapour pressure are significantly different from average profiles. Extreme surface pressure differences on the order of 80 millibars from the global average have been recorded in the past. Conditions such as these could cause an unmodelled range error on the order of 2 metres at an elevation angle of five degrees. Of greater concern are high concentrations of water vapour producing zenith wet delays of up to half a metre. Such a condition could cause an unmodelled range error on the order of 3 to 4 metres at five degrees elevation angle. Therefore, it is theoretically possible that position-critical users of wide area differential GPS, relying on models using “average” atmospheric conditions, could have their vertical position accuracy degraded by up to several metres, depending on the satellite geometry. The key question for such users is therefore: what is the frequency and magnitude of such conditions? Our research shows that significant extreme range errors are rare, provided that a good tropospheric delay model is used. We found that zenith delay errors greater than ±20 cm occurred only on the order of 7 in 100,000 cases sampled from across North America. The impact on position computation is determined by the elevation angle of the lowest satellite and whether or not a suitable weighting technique is used.