Construction of an RD model for SAR images based on an RPC model

The construction of a range-Doppler (RD) model for spaceborne synthetic aperture radar (SAR) images has traditionally relied on manual analysis of auxiliary parameter files, resulting in subjectivity and limited generalizability. An RD model is established in this research using a high-precision rational polynomial coefficients (RPC) framework. Two methods were created to determine RD model parameters and apply the RPC-based construction concept. The first is a numerical method, in which RD model parameters are derived by formulating a nonlinear equation system that integrates control-point data from the RPC model and using a least-squares iterative strategy. The second is an analytical method that applies the concept of forward intersection and establishes a step-by-step solution process based on positional information, orbit data, slant range, and Doppler measurements. The experimental analysis employed Gaofen-3 SAR images with four distinct spatial resolutions in two test sites, namely Zhongwei (Ningxia) and Songshan (Henan), China. The fitting and positioning accuracies achieved by the two methods were compared. Both RD models have a fitting error of approximately 1 pixel across four image resolutions. Additionally, the planar positioning accuracy of the two methods aligns with the RPC model, showing a discrepancy of no more than 2.6 m. These results confirm that the methods introduced performs effectively.