Ground-vehicle INS/GPS vector gravimetry assessment using repeated traverses in Montana

Abstract

We explore methods to assess the precision of gravity vector components estimated from a ground vehicle INS/GPS system. Tests of the system were conducted in western Montana with repeated traverses along several highways. The gravity estimates were obtained through a series of processing steps that included B-spline smoothing, a Kalman filter, wavelet de-noising, wave-number correlation filtering, and correction for bias and scale factor errors. The results showed significant repeatability in the vertical and horizontal gravity components estimated independently from duplicated traverses. While the estimated vertical components also agree well with existing control data (2mgal; s.d.) and in some cases had internal consistency better than 1 mgal (s.d.), the estimated horizontal components are much less coherent between repeated track s. However, after removing a scale factor error they exhibit clear correlation at all frequencies with repeatability of 1 mgal to 6 mgal (s.d.), depending on the test traverse and the horizontal component. Comparison to DEFLEC99 control data (whose validity remains in question) indicates differences up to 9 mgal (s.d.), or more.