Comparison of low angle radar clutter models

Three low angle clutter models, the compound-K, the noncentral gamma-gamma (NCGG), both of which model speckle, and the Weibull distributions are compared. The derivations of the first two densities are surprisingly similar, differing on only one critical point. The compound-K assumes that the Gaussian random variables that are the input to the square law detector have zero means, while the NCGG assumes that the Gaussian random variables have nonzero means. The compound-K must then assume that it is the output variance that is fluctuating to produce speckle, while the NCGG assumes that it is the noncentrality parameter, resulting from the addition of the nonzero means that is fluctuating. The difference in the resulting models is significant. The Weibull density has long been used to model clutter successfully, and it too will be considered for comparison, but its use is not based upon a physically motivated development. Rather, its use is based on comparisons of clutter output densities being well matched by the Weibull density with appropriate parameters.