Robust Method of Outlier Detection of GPS-Levelling Data in Sri Lanka

Abstract

The accuracy of GPS-levelling data is crucial since any errors in GPS and levelling data could cause inaccurate height determinations. In this study, outliers in GPS-levelling data were detected using Earth Gravitational Model 2008 with statistical techniques. Minimum Covariance Determinant method was used as the robust statistical technique for detecting outliers in the levelling data. It detected 35 points as outliers out of 205 observed points that dropped the standard deviation from 0.229 m to 0.083 m. It was shown that the predicted outliers were random and not consistently deviate with the rest of the points in the neighboring region.

Cite as
H.M.I. Prasanna. (2022). Robust Method of Outlier Detection of GPS-Levelling Data in Sri Lanka. Journal of Applied Mathematics and Statistical Analysis, 3(2), 1–13. https://doi.org/10.5281/zenodo.6830928