Construction of a backpropagation starlight atmospheric refraction model based on ray tracing

Abstract

The atmospheric refraction model is a vital part of the refraction navigation system. A more accurate model needs to be constructed for the navigation algorithm design and simulation verification. Based on the Thermosphere, Ionosphere, Mesosphere Energetics and Dynamics/Sounding of the Atmosphere using Broadband Emission Radiometry (TIMED/SABER) dataset and the geometrical ray propagation law, a ray tracing algorithm is proposed to calculate the atmospheric refraction angles. Moreover, a neural network starlight atmospheric refraction model (BP model) is constructed to better describe the relationship between time, location, and refraction angle. Compared with the experimental data, the bias error of the backpropagation (BP) model is 1.06′′, which is better than the 3.75′′ of the traditional model. It indicates that the BP model is exact and has important guiding significance for the starlight refraction navigation technology.