GNSS Comprehensive Stochastic Model Based on Empirical Cumulative Distribution Normalization Optimization

Objectives: In the complex landslide monitoring environment, the global navigation satellite system (GNSS) signals are affected by factors such as atmospheric delay, multipath effect, and local occlusion, resulting in the elevation or signal-to-noise ratio stochastic model based on a single index can not meet the high-precision positioning in complex scenes. At the same time, the data quality difference between stations of multi-mode and multi-frequency GNSS data in complex environment also puts forward higher requirements for the establishment of accurate and reliable stochastic model. Methods: In order to solve this problem, a GNSS comprehensive stochastic model based on empirical cumulative distribution normalization optimization is proposed. The model is constructed based on elevation and signal-to-noise ratio information, and the weight coefficient is used to refine the influence of differences between single model, multi-frequency and multi-system and stations on the stochastic model. Results: The experimental results under high occlusion environment show that the average fixed rate of the new model for ten consecutive days is 97.8%, which is 24.4%, 25.6% and 15.3% higher than that of the elevation stochastic model, the signal-tonoise ratio stochastic model and the comprehensive stochastic model with principal components analysis, respectively. The fixed solution RMS of the new model is 0.007m in E and N direction and 0.012m in U direction. Conclusions: The new model improves the epoch fixed rate while the accuracy of fixed solution can meet the requirement of GNSS centimeter positioning in complex landslide monitoring environment.