Evaluating Aircraft Positioning Methods for Airborne Gravimetry: Results from GRAV-D’s “Kinematic GPS Processing Challenge”

T.M. Damiani, A. Bilich, G.L. Mader

Abstract:	The National Geodetic Survey’s (NGS’s) Gravity for the Redefinition of the American Vertical Datum (GRAV-D) program plans to collect airborne gravity data across the entire U.S. and its holdings over the next decade. The goal is to build a geoid accurate to 1-2 cm to use as the next national vertical datum in 2022. The airborne gravity survey of the nation has been proven to be the key to accomplishing that goal. The first phase is underway, with > 26% of data collection across the U.S completed as of February 2013. To achieve the best airborne gravity data accuracy possible, the GNSS position solutions of the aircraft must provide not just accurate positions, but also accurate velocities and accelerations to be used in calculating the gravity corrections. However, no comprehensive comparison has been done of available kinematic GPS processing techniques as they pertain to producing accurate airborne gravity results. In Fall 2010, NGS issued the “Kinematic GPS Challenge” to the GPS processing community, asking for voluntary kinematic GPS processing for geodetic-quality L1/L2 GPS data on two GRAV-D airborne gravity flights done in Louisiana in 2008. All GRAV-D airborne flights are challenging for positioning, and the Louisiana flights are no exception because of their long baselines (> 500 km), high altitude (35,000 ft; 10,668 m), high speed (280 knots; 144 m/s), location (over half of each flight is over water in the Gulf of Mexico), and the need to recover accurate first and second derivatives from the final position solution. The goal of the Challenge is to directly compare position results from a variety of methods, as well as compare the gravity results computed from each of those position solutions. The flights for the Challenge were chosen particularly for this purpose. The gravity results from the Challenge will show the performance of positioning on both low-noise and high-noise data. As well, gravity recovery repeatability can be tested along the same line that was twice: once on each of the two flights. Participants in the Challenge were allowed to submit either differential or PPP position solutions, or both. For both flights, they had available 1s GPS receiver data for the aircraft, 1s data from nearby CORS stations, and 1s data from an NGS-run Ashtech GPS base station at the New Orleans airport. Eighteen position results from twelve contributors (including two from NGS) were submitted in response to Call 1 of the Challenge. After initial analysis, however, the submitted position solutions were found to be mostly dominated by a “sawtooth” pattern. After much discussion, the “sawtooth” was attributed to the Trimble receiver in the aircraft, possibly due to clock reset issues. Although some conclusions could be drawn from that first call, a clean dataset was needed to truly accomplish the goals of the Challenge. In January 2013, NGS released new data for the Challenge flights. This time, data was released from a NovAtel GPS receiver and an Applanix IMU that were also on the aircraft. The results of both the first and second calls for participation in the Kinematic GPS Challenge are kept anonymous with respect to naming names, to allow full public disclosure of the results without negatively impacting any one group’s on-going methodology development. Analyses of the Call 1 data, which included the sawtooth position pattern, showed that the submitted position solutions are somewhat different, usually by +/- 0.25 m or less in X, Y, and Z. Two outlier solutions were removed from the analysis, as they had extreme differences that appeared to be due to the receiver issues. Within the rest of the results, the differential and PPP solutions were indistinguishable from one another. All of those solutions had similar differences throughout the flight, with neither method always performing better nor worse, even when over 500 km away from the airport base station. The two position solutions that we concluded produced the best gravity results both used smoothing in their algorithms. We speculate, though could not conclude due to the “sawtooth” pattern, that the smoothing directly benefits the calculation of the derivatives and thus produces a better gravity result. An additional comparison was done with the Call 1 GPS-only position submissions and the current preferred NGS GPS+IMU solution (a loosely-coupled differential GPS+IMU solution from commercial software, Applanix’s POSPac). For the high-noise flight, the loosely coupled solution produced a significantly better gravity match with the global gravity models than any of the GPS only solution. But the difference was less significant between the two solution types (GPS+IMU versus GPS-only) for the calm weather flight. This result led directly to the decision of the GRAV-D program to make the IMU a flight-mandatory instrument. In other words, flights do not occur if the IMU is not working properly. Turbulence at altitude is poorly forecast ahead of time and so it can never be predicted when the IMU will become critical to achieving a better positioning solution. After all Call 2 position submissions are received on April 1, those analyses will be done and the results given in this presentation. The earliest submissions already show that the “sawtooth” pattern seen in the Call 1 results is definitely not present in these Call 2 results. The expectation is that the new position solutions will all agree more closely without the “sawtooth” pattern. Additionally, we hope to receive coupled GPS+IMU solutions from Call 2 of the Challenge and will include those in the comparison, given how well the IMU-coupled solution performed for the Call 1 results. Such a comprehensive, international comparison of kinematic GPS processing methods has not been done before. The additional benefit of having an application (gravimetry) with which to use the different position solutions makes this NGS GPS Challenge work unique. The benefits of such comparisons will impact both the kinematic GPS processing community as well as many remote sensing communities, especially gravimetry.
Published in:	Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013) September 16 - 20, 2013 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	3489 - 3507
Cite this article:	Damiani, T.M., Bilich, A., Mader, G.L., "Evaluating Aircraft Positioning Methods for Airborne Gravimetry: Results from GRAV-D’s “Kinematic GPS Processing Challenge”," Proceedings of the 26th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2013), Nashville, TN, September 2013, pp. 3489-3507.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In