Stochastic modeling the vertical deflection errors of EGM2008 for INS/GNSS integration

The vertical deflections (VD) error of earth gravity model, which is one of primary error sources in high-precision inertial navigation system, is usually modeled as stochastic processes. In this paper, the VD error of the Earth Gravity Model 2008 (EGM2008) are modeled as 2ed order Gauss-Markov processes. The statistical characteristics of VD error are numerically analyzed by extensive comparing with a high-precision VD model of USA, namely DEFLEC12. Second order auto-regression processes are adopted to model the discrete VD error data on different trajectories, and the parameters of the corresponding 2ed order Gauss-Markov continued model are calculated. According to the statistical analysis, the general rules for setting model parameters are concluded.