Analytical and Experimental Observations of Ionospheric and Tropospheric Decorrelation Effects for Differential Satellite Navigation during Precision Approach

The introduction of the Global Positioning System (GPS) ushered in a new era of affordable precise navigation. Differential GPS (DGPS) corrections have reduced positioning errors from about 100 meters to roughly 1 meter, which has led to their proposed use in aircraft precision approach. However, the accuracy of differential corrections is somewhat limited by spatial and temporal decorrelation of the atmosphere. Current requirements for precision landing, which limit integrity risk to one undetected navigation hazard in a billion trials, suggest that we must fully characterize ionospheric and tropospheric decorrelation. This paper will present both theoretical and experimental results in this regard. Currently assumed values of spatial decorrelation for the ionosphere and troposphere are about 2 and 1 mm/km respectively. This suggests an error of only 2 centimeters error due to the ionosphere, and 1 centimeter error due to the troposphere at 10 kilometers from the DGPS reference station. However, these are simply average values based on relatively smooth models of the ionosphere and troposphere. These values seem overly optimistic when compared to a documented case of ionospheric decorrelation that resulted in a 50 centimeter difference between ionospheric estimates over a 9 kilometer baseline. Models of the troposphere were also analyzed to examine the variability due to changing surface conditions. Archived inflight data from a prototype Local Area Augmentation System (LAAS) architecture was post processed using several different tropospheric models. Carrier phase residuals were calculated at each epoch as a measure of model consistency. The actual displacement due to a change in model was examined to determine the sensitivity to ground measurements of temperature, pressure and humidity. The archived LAAS flight data was input to a Carrier Smoothed Code algorithm, along with an assumed ionospheric gradient. The impact of filter time constants on pseudorange error was calculated. This paper seeks to obtain statistical bounds for the tropospheric and ionospheric decorrelation effects on a LAAS architecture. Introduction Measurement decorrelation is an issue for precision DGPS applications such as LAAS. Decorrelation is roughly linear with range from the reference station, thus it becomes less of a threat as you move farther from (and higher above) the airport. Previous GPS research in Antarctica measured a decorrelation of 50 cm over a 9 km baseline, all of which was attributed to the ionosphere [Goad]. This is equal to a decorrelation gradient of 55 mm/km, which is 20 times larger than the typical quoted values. While polar regions typically experience larger values of TEC, this single observation is still quite alarming since it is certainly not the worst possible point, just the worst one that was observed during the short campaign. The dry component of tropospheric delay constitutes about 90% of the tropospheric delay and is fairly stable over periods of an hour or so. The wet component is smaller, but is much less stable changing considerably over the course of an hour or during a displacement of several kilometers. The first half of this paper will address the tropospheric concerns, while the latter half focuses on the ionosphere. MASPS Requirements The RTCA has derived upper bounds for the residual errors due to DGPS for the Signal in Space (SIS). This model is given by the following equation: