Improving long-range RTK: getting a better handle on the biases

GPS World, March, 2008 by Don Kim, Richard B. Langley

Biases and errors such as satellite orbit error and atmospheric signal refraction are the primary limiting factors in successful long-baseline, real-time kinematic (RTK) style processing of GPS measurements--either in real-time or post-processing mode. These error sources are dependent on the distance between a reference and rover receivers. If they are not adequately accounted for, they can result in significant positioning errors in long-baseline applications. This is particularly true for the conventional single-baseline RTK and hence reduces the effective inter-receiver distance of this technique to a few tens of kilometers.

We can apply effective strategies to mitigate these error sources. For example, the ionosphere-free linear combination of the L1 and L2 carrier-phase measurements can completely cancel first-order ionospheric delays. Although this approach is appealing for mitigating the ionospheric errors, we have to be prepared to accept some penalty. As it is difficult to fix integer ambiguities using the ionosphere-free observations for long baselines, float ambiguity solutions (less accurate than fixed ones) are normally used. Due to the amplification of the noise by the linear combination, the solutions are less precise. Errors in broadcast GPS satellite orbits have little effect for baselines up to a few hundred kilometers and, furthermore, can be virtually eliminated using precise ephemerides in post-processing mode. Tropospheric delay is usually estimated based on model atmospheric predictions and/or surface meteorological observations made near the stations at the time of the GPS measurements. As this approach often inappropriately accounts for spatial and temporal variations in water vapor delays, it is a common procedure to estimate a residual zenith delay from the data itself.

As an alternative approach to mitigating the error sources, network RTK based on multiple reference stations is often used. The integration of several reference stations into a combined network provides a capability for modeling the error sources at a rover within the network and enables lengthening the baselines up to a few hundreds of kilometers. Despite successful implementation of network RTK for long-baseline applications, however, its performance is not always equivalent to single-baseline RTK operating in short-baseline situations. As network RTK interpolates error corrections for a rover using the error estimates at reference stations, this approach is vulnerable to localized anomalous errors under unfavorable atmospheric conditions. For example, weather fronts and atmospheric conditions associated with heavy rainfall can cause rapid variations in the tropospheric delay and, subsequently, the performance of an RTK system can be significantly degraded even across relatively short baselines. Such anomalies are not canceled in the interpolation procedure used for deriving rover delays. Also, solar-terrestrial interactions can cause significant changes in the morphology of the ionosphere, changing the propagation delay of GPS signals within time intervals as short as one minute. Such changes can last for several hours primarily in the polar, auroral, and equatorial ionospheres. During severe ionospheric activity, the correction accuracy deteriorates and adversely affects the ambiguity resolution over the network. When a rover is located outside the network boundary, network RTK must extrapolate error corrections for the rover. As a result, network RTK can face the same challenges as single-baseline RTK.

Over the past few years, University of New Brunswick (UNB) researchers have carried out several projects involving long baselines that, unfortunately, could not take advantage of network RTK. These included a field experiment to investigate the performance of different neutral atmosphere mitigation strategies during the 2005 mission of the Canadian Coast Guard Ship Amundsen (a research icebreaker) in the Canadian Arctic and Hudson Bay, and collaboration with the University of Southern Mississippi to advance positioning results by means of improved differential tropospheric modeling in the marine environment of the Bay of Fundy in eastern Canada. In both studies, the number of reference stations deployed was not sufficient to adequately model the errors using network RTK. Instead, our approach for achieving high accuracies at greater distances from differential reference stations was to use single-baseline RTK in a novel post-processing mode.

In this article, we describe our new approach for long-range RTK. Although this approach was originally developed for single-baseline RTK over long distances in kinematic mode, it can be used for network RTK when requiring extrapolation of the differential ionosphere corrections for a rover located outside the network. It can also be used in cases where the rover located inside the network is experiencing local anomalies in the differential ionospheric delays.

New Approach Considerations