Log-Domain Statistical Modeling of S-Band Rain Clutter for Surveillance Radar Applications

Rain clutter degrades surveillance radar perfor- mance by masking weak targets and increasing false alarms. Classical clutter models are formulated in the linear-amplitude domain, yet the large dynamic range of rain returns motivates a logarithmic transformation for more tractable statistical characterization. This paper analyzes the statistical characteristics of log-transformed rain clutter using 372 samples from an S-band surveillance radar dataset. Six candidate distributions (Log-Gamma, Log-Nakagami, Log-Rice, Log-Rayleigh, Log-Weibull, and Log-K) are fitted via maximum likelihood estimation and compared using the Akaike Information Criterion (AIC), the Bayesian Information Criterion (BIC), and the Kolmogorov-Smirnov and Cramér-von Mises criteria. The results support log-domain modeling for improved CFAR detection and motivate the development of a compound distribution to provide more precise modeling for S-band rain clutter, an approach that can be further extrapolated to other frequency bands.