Abstract In GPS short-distance static relative positioning (e.g. the length of the baseline shorter than several kilometers and its height difference less than tens of meters), the precision and accuracy of the baseline solutions are limited because of the existence of unmodeled errors especially multipath errors. Considering that multipath errors are mapped from original observations to the baseline solution (position domain) and its residual series (measurement domain), scholars have applied a number of methods to reduce multipath errors in residual series to improve the precision of the baseline solution. However, multipath errors are difficult to eliminate from the baseline solution. Therefore, this study proposes a new strategy for eliminating the multipath errors both in the position and measurement domains. The steps are as follows: (1) a post-processed kinematic (PPK) approach was used to obtain the coordinate and residual series with L1 observations; (2) the empirical mode decomposition (EMD) was employed to extract multipath errors from the residual series to correct original observations; (3) PPK was used again to obtain the corrected coordinate series with the corrected observations; (4) finally, a multi-scale decomposition utilizing wavelet transform was applied to the corrected coordinate series to obtain the residual terms and their average values, the result of which is taken as the final baseline solution. The precision and accuracy of the baseline solution are evaluated according to the root mean square error (RMSE). The experimental results show that the accuracy of baseline solution are improved about 16, 24, and 28 % for 10 min time periods, respectively, 24, 25, and 33 % for 30 min time periods, respectively, and 5, 23, and 22 % for 60 min time periods, respectively, of the traditional scheme. The influence of multipath errors on the baseline solution are efficiently weakened.
"Acta Geodaetica et Geophysica" – Springer Journals
Published: Mar 1, 2016