Influences of different orbit interpolation methods on substitution accuracy of rational polynomial coefficient model for multi-mode images from Gaofen-3 Satellite

Abstract

Determining the satellite orbit vector is necessary when constructing the geometric positioning model for a Synthetic Aperture Radar image from the Gaofen-3 satellite (GF-3), as it greatly impacts the geometric positioning accuracy. Therefore, it is vital to obtain accurate orbit vector data regarding the satellite imaging time. Here, GF-3’s orbit was interpolated using the Lagrange interpolation, Chebyshev polynomial, and ordinary polynomial methods, with each method’s influence on the substitution accuracy of the Rational Polynomial Coefficient (RPC) model being analyzed for GF-3’s various imaging modes. The results show that based on Lagrange interpolation orbit, the accuracy of RPC substitution is greatly affected by the length of the orbit, and the stability of RPC substitution accuracy is limited by the position of the interpolation orbit segment. In general, this method shows low RPC substitution accuracy and large fluctuations. The Chebyshev polynomial method and the ordinary polynomial method are less affected by the orbital length and can obtain high substitution accuracy. The RPC substitution accuracy of the two methods was higher than 0.08 % and 0.02 %, respectively. In addition, the results of RPC substitution accuracy are more stable and reliable when ordinary polynomial interpolation orbit is used.