On the use of finite surfaces in the numerical prediction of rough surfacescattering

Method of moments (MOM)-based Monte Carlo calculations are widely used in d etermining the average radar cross section of randomly rough surfaces. It i s desirable in these numerical calculations to truncate the scattering surf ace into as short a length as possible to minimize the solution time. Howev er, truncating the surface tends to change the solution for the surface fie lds near the truncation points and may alter the scattered far fields. In t his paper, these end effect errors are examined for one-dimensional (i.e., grooved or corduroy) surfaces which are Gaussian distributed in height and have either a Gaussian or a Pierson-Moskowitz spectra, In the case of the P ierson-Moskowitz type surface, it is shown that a relatively short surface of 80-120 wavelengths can be used to obtain the average backscattered radar cross section for backscattering angles as large as 60 degrees from normal . For a comparatively smooth Gaussian surface, on the other hand, its is sh own that the truncation effects can be very significant at moderate backsca ttering angles. Also, great care should be taken when examining the scatter ing from Gaussian surfaces which are dominated by specular scattering. It i s shown that in this situation, a very large number of calculations may be needed to obtain a good numerical average.