Analyzing the Statistical Characteristics of Clutter from the Sea Surface

Abstract

As is known, when radars with long impulses observe a sea surface area in a range much larger than the seawave length, the clutter in the form of reflections from the sea surface has the Gaussian distribution, and their envelope has the Rayleigh distribution. As the impulse width decreases (i.e., as the radar resolution range is increased along with significant influence of small grazing angles), the clutter distribution differs from the Gaussian one, and the envelope distribution law differs from the Rayleigh distribution in having a longer "tail". The clutter becomes impulsive in nature. Thus, the radar receiver designed proceeding from the standard assumption that the sea clutter obeys the Gaussian distribution becomes less effective, because the false alarm probability is increased due to an incorrectly chosen object detection threshold that is not designed for the impulse nature of clutter. Hence, a need arises to choose the most adequate model of disturbing sea clutter. The article presents a review and analysis of the main models of non-Gaussian sea clutter (lognormal, Weibull and K-distribution) that are used to describe disturbing sea clutter. Much attention is paid to the model based on the K-distribution. The experimental records of sea clutter obtained using the IPIX coherent-pulse radar were compared with those obtained using the main sea clutter models. The study results have shown that the lognormal model tends to overestimate the dynamic range of the clutter actual distribution, which means that the distribution has a longer "tail". The Rayleigh model tends to underestimate the dynamic range; that is, the tail is significantly shorter. The clutter model based on the Weibull distribution offers much broader possibilities for statistically modeling sea clutter than those offered by the lognormal model or the Rayleigh model. Depending on the parameters, the Weibull distribution can transform into the Rayleigh or lognormal distribution, thus reflecting the distribution of real clutter in a more accurate manner. In comparison with the other considered models, the sea clutter model based on the K-distribution describes the sea clutter envelope structure most accurately, even in the zone of distribution "tails". Thus, the clutter model based on the K-distribution is the most promising one for describing the sea clutter envelope in a radar with short direct impulses.