Ambiguity Resolution in the Air

The need for ambiguity resolution in aerial survey limits aircraft banking angle to a flat turn. It also inhibits flying far from reference stations. Technology is now in place to eliminate these hurdles.<P>

Ambiguity resolution is the search for the correct number of integer cycles of the L1 carrier signal between rover and each satellite. Estimation of ambiguities requires a continual lock onto the signal from each satellite. Turning the aircraft into a steep banking angle, say larger than 15 to 20 degrees, can result in the wing blocking view of the satellites, causing resetting of the solution.

SmartBase
Flying a turn at a 15-degree banking angle at a speed of 150 knots gives a turn rate of about 2 degrees per second, and turn radius of 2.3km; it will take one minute to complete a 180-degree turn. In contrast, banking at 30 degrees will allow the same turn to be made in half the time and radius, saving time, reducing fuel costs, or allowing more lines per aerial survey. A smaller radius also allows more flexibility in restricted airspace where room for safe manoeuvring may be sparse. SmartBase and IN-Fusion technology implemented in POSPac MMS enables aerial survey to be flown at banking angles above 20 degrees. Based upon Trimble VRS technology, SmartBase is optimised for large changes in altitude and extended to work with reference stations separated over large distances. Initial ambiguity resolution requires only that the aircraft be within the network at a maximum distance of 70km from a reference station; after initialisation the aircraft may move up to 100km from a reference station while accuracy remains at 10-15cm RMSE level.

Limitations Solved by IAKAR
More than 30km from a reference station, atmospheric delays mean that ambiguities cannot be reliably estimated. Hence traditional Kinematic Ambiguity Resolution (KAR) algorithms require the aircraft to be within 30km of a reference station for a certain time to obtain ambiguity resolution. Once ambiguity resolution has been established, the aircraft should not move farther than 75km from the reference station or the error will become too big. These limitations are solved by Inertially-Aided Kinematic Ambiguity Resolution (IAKAR) implemented in IN-Fusion technology, which processes inertial data and raw GNSS observables (phase and range measurements) in a single, tightly integrated Kalman filter. In the case of cycle slip or outage in the GNSS data, the inertial data allows immediate re-establishment of the ambiguity. The need to fly flat turns is eliminated.

Virtual Reference Station
On land, productivity of Real-Time Kinematic (RTK) positioning has been improved with the concept of Virtual Reference Station (VRS); observables from GNSS reference stations are pro­cessed to compute the atmospheric and other errors in the network. These are then interpolated to generate GNSS observations as if a reference station was located at the rover. Benefits include the following:- nearest reference station can be further away than 30km
- integer ambiguities can be more quickly fixed and overall reliability increases
- no need to set up base-stations, so saving costs
- in contrast to a centralised multi-base approach, no special processing is required in the RTK engine.

VRS enables real-time accuracy at centimetre level (RMSE) anywhere in the network. Rigorous adjustment of all station positions over an eighteen to twenty-hour period to ensure correct and consistent positioning is routinely done in land surveying as part of best practice, but has been a weak point in aerial survey. Too often, data from a single reference station or CORS (continuously operating reference system of the USA National Geodetic Survey) is used without proper quality control. Failure may concern published antenna coordinates, datum or observables. SmartBase includes such rigorous adjustment. Tests show that better than decimetre-level (RMSE) can be achieved with a sparse network of four reference stations separ­ated by over 100km, although results depend greatly upon the dataset. Robustness is greatly improved for networks such as CORS, the Japanese Geographical Survey Institute (GSI), or German State Survey Satellite Positioning Service (SAPOS) consisting of over 250 permanently observing, nationally extending reference stations. The area that can be flown is virtually limitless.

Tests
Using data from a Lidar calibration flight collected in April 2007, BLOM Geomatics AS of Norway investigated the accur­acy of POSPac MMS and the potential for flying sharp turns. From 27 reference stations located throughout Norway, Denmark and Sweden, four separate SmartBase networks were created with baselines of respectively 60km, 110km, 200km and 300km. The solutions for each network were compared with a reference, and positional differences for the 60, 110, and 200km networks were all well below 10cm; orientation differences were well below 30 arcsec. Only for the 300km network did the differences start to grow as the software had trouble resolving the ambiguities. Outages due to high banking turns were simulated by turning off the raw GPS observations at the centre of the trajectory so that only two satellites were tracked; in one case for 30 seconds and in the second for 75 seconds. In both cases the positional differences remained below 10cm and were virtually unaffected before and after outages. Performance results will further improve as additional GNSS observables are added.

Joe Hutton, Applanix Corporation, Director, Airborne Products, 85 Leek Crescent, Richmond Hill, Ontario, Canada L4B 3B3, email: jhutton@applanix.com