GPS Levelling Results from Two Test Areas in Finland

Abstract

GPS levelling was used in two test areas in Finland in order to investigate the capability of GPS measurements for orthometric height determinations. Both test areas are surrounded by the spirit levelling loops of the Precise Levelling network. Besides, the test areas are covered by a dense network of lower order levelling lines that are connected to the Precise Levelling network. The GPS networks measured in both areas consisted of 51 and 45 GPS sites that were located at spirit levelling benchmarks. The average side length was 15 km.

Four different geoid models, two global models (0SU91A, EGM96) and two local models (NKG89, FIN95), were used to convert the GPS ellipsoidal heights to orthometric heights, after which the results were compared to the spirit levelled heights. In all cases the height differences showed a clear tilt which was removed by fitting first and second order surfaces to the differences. When the global geoid models were used for the conversion, the RMS of the height differences varied from ±40 mm to ±80 mm. The RMS of the differences was reduced to appr. ±15 mm when the local geoid models were used. When the errors caused by the spirit levelling in the computed height differences were removed, the estimation of the average accuracy of the GPS levelling on the test areas gave the result ±12 mm.