Wave optics simulation of statistically rough surface scatter

The bidirectional reflectance distribution function (BRDF) describes optical scatter from surfaces by relating the incident irradiance to the exiting radiance over the entire hemisphere. Laboratory verification of BRDF models and experimentally populated BRDF databases are hampered by sparsity of monochromatic sources and ability to statistically control the surface features. Numerical methods are able to control surface features, have wavelength agility, and via Fourier methods of wave propagation, may be used to fill the knowledge gap. Monte-Carlo techniques, adapted from turbulence simulations, generate Gaussian distributed and correlated surfaces with an area of 1 cm2 , RMS surface height of 2.5 μm, and correlation length of 100 μm. The surface is centered inside a Kirchhoff absorbing boundary with an area of 16 cm2 to prevent wrap around aliasing in the far field. These surfaces are uniformly illuminated at normal incidence with a unit amplitude plane-wave varying in wavelength from 3 μm to 5 μm. The resultant scatter is propagated to a detector in the far field utilizing multi-step Fresnel Convolution and observed at angles from −2 μrad to 2 μrad. The far field scatter is compared to both a physical wave optics BRDF model (Modified Beckmann Kirchhoff) and two microfacet BRDF Models (Priest, and Cook-Torrance). Modified Beckmann Kirchhoff, which accounts for diffraction, is consistent with simulated scatter for multiple wavelengths for RMS surface heights greater than λ/2. The microfacet models, which assume geometric optics, are less consistent across wavelengths. Both model types over predict far field scatter width for RMS surface heights less than λ/2.