Fast simulation of a moving sea surface remotely sensed by radar

This paper describes the fast simulation of a sea surface remotely sensed by radar, by using parallel computing on GPU cards. The approach which consists in adding two “grids” of the surface with different spatial resolutions is tested. The high-frequency (HF) grid corresponds to the HF part of the resolved surface spectrum, and the low-frequency (LF) grid to its LF counterpart. Numerical tests on the surface slope PDF and autocorrelation function show that adding these two grids with interpolation of the LF grid give very satisfactory results, provided that an appropriate windowing is applied. Then, this very efficient surface generation makes it possible to simulate a moving sea surface remotely sensed by radar with highly reduced computing resources.