Fast Algorithms for Imaging Simulation through Turbulence

Current techniques for simulating light propagation through atmospheric turbulence employ Fourier transform methods for calculation of diffraction effects. Although highly accurate, these methods are computationally burdensome, and substantially slow the task of computing extended scenes on a time-varying basis, as is required for closed-loop analysis of laser weapons systems' performance. In other areas of computational wave simulation (fluids and electromagnetics) Fourier techniques have been replaced by specialized finite-difference techniques which offer comparable accuracy for substantially reduced computational cost. Alternatively, these same diffraction effects seem amenable to local solution by integral techniques which would also provide a speed-up over use of Fourier methods. Potential speed-ups for the diffraction portion of the simulations range from 6 to over 150. Both alternative methods handle general boundary conditions much more gracefully than Fourier techniques. LRK Associates proposes to adapt both the finite-difference and integral techniques to the scene propagation problem, testing their accuracy and measuring their computational advantages on some simple test problems. This will lead to recommendations on which techniques are the best candidates for implementation in the Phase II to replace Fourier transform techniques for this optical simulation application.