Comparison of uncertainty analysis methods for laser horizontal propagation in the atmosphere

Laser beams are affected by diffraction, turbulence, and thermal blooming effects when propagating in the atmosphere, leading to significant uncertainty in the far-field 63.2% encircled average power density (Ien) at the target. Focusing on key uncertainty sources (jitter, initial beam quality factor, temperature, wind speed, C2n, absorption coefficient, and extinction coefficient) of Gaussian beams for horizontal propagation in the atmosphere, this paper investigates the uncertainty of Ien. Employing the constructed scaling law model of Ien as a fast computational tool, the results of three uncertainty analysis methods are compared, including Latin Hypercube Sampling (LHS), arbitrary Polynomial Chaos Expansion (aPCE), and generalized Polynomial Chaos Expansion (gPCE). The results show that the three methods produce consistent estimates of the Ien mean, standard deviation and relative standard deviation. However, there are significant difference in computational efficiency: the LHS benchmark requires 12000 model evaluations, whereas gPCE and aPCE just need 4000 samples, beyond which additional samples have negligible influence on the output. Furthermore, aPCE and gPCE reach convergence at expansion orders of 2 and 3, respectively, demonstrating that aPCE converges more rapidly in the propagation scenario of this study. This research provides practical insights for selecting uncertainty analysis methods during design and operational application of laser systems.