GPS altitude - results from analysis
of a sample of IGC flight data files
by Ian Strachan, Chairman IGC GFA Committee
What follows is based on para 7 of the annual report from the GFAC Chairman to the IGC Plenary meeting that was held on 9-10 March 2001. The full agenda was available through: http://www.fai.org/gliding/meetings
7. GPS Accuracy - Withdrawal of SA - GFAC Tests. On 1 May 2000 the Selective Availability (SA) error that applied to public domain receivers of the US GPS system, was removed. Based on tests made by GFAC, the average accuracy of lat/long fixes recorded in IGC data files improved from 44 metres (36m with 12 channel receivers) with SA on, to 12.9 metres with SA off (12 channel receivers). These tests were made in a ground vehicle with respect to an accurately-surveyed ground point at about 51N 001W. (2004 update: with an increased sample size, the average lat/long error using IGC data files from GFAC tests has reduced to 11.5m)
7.1. GPS altitude recorded in IGC data files. After the withdrawal of SA, suggestions were immediately made, particularly on the Gliding Internet Newsgroup (r.a.s.), that GPS altitude was now accurate enough to replace pressure altitude for Sporting Code and other purposes. It was felt that this should be confirmed by analysing a sample of IGC flight data files. As a result, over 400 IGC files from glider flights after 1 May 2000 were analysed. These were from 9 countries, mainly from flights on day one in 15 different competition classes.
7.1.1. Expected results. In about three-quarters of the files, results were as expected with a good comparison between the shape of GPS and pressure altitude graphs with time. Due to the different position-line geometry for altitude fixes compared to lat/long, an average error of roughly twice that for lat/long would be expected (about 26m based on a lat/long error of 12.9m).
7.1.2 Poor results. However, in about 27% (111) of the files, results were unexpectedly poor with an average variation of 211m (692ft) on the expected difference between GPS and pressure altitudes. It is emphasised that this is not a reflection on the inherent altitude accuracy of the GPS system itself, but on how it is recorded in current IGC-approved flight recorders in actual sport gliding flights. It also does not mean that GPS altitude cannot be used in the future for accurate IGC altitude recording, perhaps after enhancements to the Specification for GNSS Recorders and more rigorous altitude testing before IGC-approval is given. Future improvements to the GPS system are being monitored such as the new L5 frequency that should further improve accuracy. See also para A6 below on conversions. The results are reported in more detail in the Appendix that follows.
7.1.3. Recommendations. In view of 7.1.2 above, no change to the current Sporting Code rules on the use of pressure and GPS altitude can be made at this time. However, pilots are recommended to check GPS altitude results from IGC files and ensure that the glider GPS antenna and wiring is undamaged and is mounted in a good position in the glider. Analysis continues and discussion is taking place with some manufacturers of GNSS Recorders and GPS receiver boards. It should be borne in mind that pressure altitude used for IGC altitude achievements (and aircraft Flight Levels) has a different vertical scaling to GPS altitude, and a different lower datum. See the Appendix para A6 for more details.
APPENDIX: IGC FILES - ALTITUDE RESULTS AND ANALYSIS
A1. GPS altitude - consistency in IGC data file records. Para 7.1 above gives the background. We expect the GPS and baro altitude records ("traces") with time to be reasonably parallel (but not the same) over a small range of time and altitude. However, in 111 out of 410 (27.1%) IGC files analysed, there was an average inconsistency or variation of 211m (692ft) from the expected parallel traces, particularly for low and high points that would be used for gain-of-height claims. An example is given at the end of this appendix. These variation figures were recorded on an Excel spreadsheet so that analysis could be carried out. There were also cases where GPS altitude was not recorded at all for a short time ("altitude drop-out"), 10 or more drop -outs being shown in 28 of the IGC files analysed (6.8%) and 5 or more in 44 files (10.7%). In most of these cases, lat/long fixing appeared to be unaffected.
A2. Altitude results. The poor results from the 27% of IGC files mentioned above were unexpected and disappointing. This could have been due to the antenna installations in the gliders concerned. Also, some types of GPS receiver boards performed better than others, but in all cases of poor altitude results, good results were found in IGC files from the same type of recorder in other gliders. The situation has probably been obscured in the past by the fact that Lat/long fixes appeared to be normal and consistent with adjacent fixes which had good GPS altitude, so that validations of Observation Zones were not affected. The difficulty for sport aviation generally is the feasibility or otherwise of technical inspection of individual installations to ensure optimum GNSS system performance including antenna and cabling quality and position. An interim report to this effect was made to the FAI General Sporting Commission (CASI) by Ian Strachan (UK CASI delegate and CASI Secretary) and Tor Johannessen (IGC Delegate to CASI) at their meeting in Sweden in September 2000 and can be seen as Annex E to the CASI minutes through www.fai.org.
A3. Recommendations to pilots. Pilots should look carefully at GPS altitude traces as well as those for lat/long. If any anomalies are noted, the glider installation should be checked, particularly the antenna position, connections and cables. Glider structure or other equipment which may obscure GPS signals should not be above or to the side of the antenna, which needs a clear radio horizon for best effect. It is believed that some installations were behind the pilot in the stowage between the wings, in some cases in gliders where carbon fibre is used above this compartment. There were also some cases where the antenna was disturbed by stowage of other equipment.
A4. Earlier GFAC tests. Anomalies such as altitude drop-outs have not been found in tests that are made by GFAC for initial IGC-approvals. These are generally conducted under conditions of good signal strength.
A5. Example of altitude variations. Assume that the altitude flight records for a section of an IGC file show an average difference between GPS and baro as 100 metres. If this difference with each fix is nearly constant at about 100m, there is no inconsistency. This is what we would expect as long as the range of altitude during that section of the flight is relatively low. However, in the 27% of files mentioned, there were significant "plus and minus" variations. For instance, in one fix the baro/GPS difference may have been recorded as 270m, and in another 60m, instead of the 100m expected in this example. This is a variation of +170m (270-100) and -40m (60-100), a total variation of 210m (170+40), the average for the 27% of IGC files mentioned. On the spreadsheet, variation figures under 100m were not counted in the 27%. Also, expected variations due to large altitude range were not counted, the altitudes for the plus and minus figures in the 27% differing by an average of only 523m.
A6. Pressure and GPS altitude scales and conversions. Pressure altitude used by IGC (and aviation generally) is taken from the ICAO International Standard Atmosphere (ISA) that tabulates notional altitudes against pressure, density, temperature, and so forth. Also, IGC altitude achievements and aviation Flight Levels use a sea level datum of 1013.25 mb at 15 deg C. GPS altitude is vertical geometric distance calculated from either an Ellipsoid or Geoid datum. It is therefore expected that there will be differences between the GPS and pressure altitude measurements unless by co-incidence the conditions on the day are such that the values are similar at a given altitude and time. The ellipsoid datum is straightforward and is the ellipsoid relevant to the selected Geodetic Datum (see the Sporting Code Glossary under "Geodetic Datum"). Some GPS boards include a setting for GPS altitude labelled "sea level" or similar. This is achieved through an electronic look-up table for the relevant Geoid, which is a theoretical world surface of equal gravitational potential, approximately equal to local sea levels. The maximum differences between the WGS ellipsoid and the WGS Geoid are +65m at 60N 030W (S of Iceland) and -102m on the equator at 080E (S of India). Accurate conversion of Geometric to ICAO Pressure Altitude and vice-versa involves formulas which include details of the local atmospheric structure for the day, such as pressure, temperature and humidity at sea or ground level and with altitude. The FAI Sporting Code Section 2 (Ballooning) contains some information on such conversions (in its Annex 2), and this and other sources are being investigated in case this could be useful to IGC in the future.
-------- extract ends --------
Excel data: File Acc-GPSalt-Long.xlw
