The Kinematic GPS Challenge: Supporting Airborne Gravimetry Missions
November 29, 2011 By: Gerald L. MaderThe National Geodetic Survey (NGS) is conducting a 12-year project to re-define the vertical datum of the US. This project, referred to as GRAV-D (Gravity for the Redefinition of the American Vertical Datum), is accomplished by flying airborne gravity missions over the continental and coastal US. Gravity solutions require accurate sensor positioning and accelerations, which in turn requires precise and accurate GPS solutions for the aircraft.
To help us facilitate our software and method development, we invited interested researchers and companies to compute and submit solutions from samples of actual GRAV-D data. A number of kinematic techniques and methods suited for long distance flights have emerged over the past few years. Given the challenging nature of these flights (long distances, high altitudes, varying tropospheric regimes) and the importance to GRAV-D, we believe applying these techniques to such a common data set would also be of great interest to the kinematic community.
The following files were made available for each of the participants to download.
- Two flights were included in the data set: days 08_297 and 08_324. RINEX files were included for the airplane (mgps), the base station at the airport (0009) and several other nearby CORS. All data were at a 1-sec rate.
- A file (BaseStationInfo) containing base station ARP xyz’s and antenna types was included to ensure all participants used the same station metadata.
- The aircraft was a NOAA Citation. A relative antenna calibration had been performed for this aircraft and its block of data was included for insertion into the antenna calibration file.
- The RINEX Nav files and IGS precise orbit files were also provided.
The flights were in Louisiana, perpendicular to the coast and were about 500 km long, with about half of the flight over land and half over the Gulf of Mexico. The typical flight characteristics for a GRAV-D mission are: Altitude = 35,000 ft, ground speed = 280-320 knots (144-165 m/s), flight length 3-4 hours. A Google Map showing the 2 flight trajectories and the available base stations is shown in Figure 1.
Participants were asked to provide their solutions for the 2 flights expressed as xyz coordinates (ITRF) of the aircraft antenna’s ARP for each second of the flight in GPS time. NGS received 17 solutions from 11 different participants which included differential as well as PPP solutions from a variety of software packages.
Figure 1. The trajectories of the two kinematic GPS test flights. The airport was at the station marked 0009 and the other CORS were available for use.
The flight profile for the flight on 08-297 is shown in Figure 2. The times when the aircraft was on the ground at the airport, ascending or descending, and in operational flight, along with it’s distance from the airport is shown. Although not explicitly shown, the flight profile for the second flight, 08-324, is similar. The comparisons are shown for the vertical component of the 2 flights in Figures 3 and 4. For simplicity, 1 of the solutions was selected as a reference allowing the differences of each of the remaining solutions to be shown with respect to this solution.
Figure 2. The aircraft altitude and distance from the airport for the first flight on 08_297 is shown. The brief stationary portion at the airport at the beginning and end of the flight is easily distinguishable along with the ascending and descending portions.
Figure 3. The up-differences for day 08_297 of each of the participating kinematic solutions with respect to one of the solutions, chosen arbitrarily.
Figure 4. Similar to Figure 3 but for day 08_324.
The flight on day 08_297 was more turbulent than usually desired. For that reason it was reflown on day 08_324. Given the same basic flight profile and the difference in flying conditions, this pair of flights was selected for its additional interest to the GRAV-D project. It is clear from the appearance of Figures 3 and 4 that the difference in turbulence had a significant effect on a number of the solutions. Ideally, we hoped to see the solution overlap each other within a narrow band. Since the airplane flew at an altitude that exceeded by far any previous experience the participants had, we expected differences in the tropospheric modeling to show up as more-or-less constant offsets between the various vertical solutions. Despite the clutter, this can be seen for a number of the solutions on both days, but especially for the calm day (08_324).
The differences seen while the plane is rapidly ascending and descending are not too surprising. However, the discrepancies while the airplane was at rest on the ground are particularly curious. Excepting one outlier at the beginning of the flight on day 08_324, the up-components are spread over 20cm at both the beginning and end. The spread on day 08_297 is twice as large.
Eventually these results are turned into accelerations by taking the second derivative of these vertical components. These too resulted in significant differences among the solutions which will be reported later (Theresa Diehl, AGU, December 2011). Given the sub-centimeter agreement of static positioning from most of these same participants using the same or similar software, these differences indicate that there is considerable room for improvement in our kinematic GPS processing. We hope that we may continue these comparisons on different data sets with the aim of reconciling these differences.
RSS
Become a Fan
Join our Network
Follow Us
Please send any technical comments or questions to our webmaster.