Goodness-of-fit Based Weather Radar Ground Clutter Model Selection

In many studies, different models for weather radar clutter signal are used, each with its advantages and disadvantages. In this paper we use a model selection method through a goodness-of-fit (GoF) test over its power spectral density. The Barlett's T_p test's performance is firstly studied using synthetic radar data. This specific test has the advantage of being model independent. Using this test we compare the GoF of different clutter models to real measurement data obtained from an Argentinian weather radar (Radar Meteorológico Argentino, RMA). The Gaussian shape for the Power Spectral Density (PSD), both with and without considering windowing effects, and a first order autoregressive (AR) model are evaluated, since they are the most popular in weather radar applications. We also suggest truncating the spectrum to the clutter mode because it shows an improvement for the model selection. As a result, the first order AR model offers a higher rate of test acceptance than the other models.