Study of the modeling of radar sea clutter using the KA distribution and methods for estimating its parameters

Effective threshold signal detection in airborne radar systems operating above sea requires an accurate and tractable statistical modeling of the scattered returns from the sea surface, namely the sea clutter. When high resolution radars operate at small grazing angles, the statistics of the signal-generated clutter are highly non-Gaussian. When this is not taken into account, it leads to poor signal detection performance. This paper emphasises on the statistical description of the sea clutter using the KA distribution. The description includes the ambiant noise as well as the effect of the wave surface structure. The KA distribution lies on an appropriate combination of the Class A model, established by Middleton, and the description of the wave structure through the Gamma distribution. The use of the KA distribution as an accurate and tractable model for the scattered returns from the sea surface, embedded in the ambiant noise, has been established in an article by Middleton and in another one by Ward. However, the probability distributions of the two articles exhibit some differences on which the first issue of this paper is to shed light. The second issue of this paper is to illustrate that the KA distribution of Middleton, as it is now distinguished from Ward´s, shows an excellent agreement with real experiments. This is done using airborne radar records in rough sea conditions. To estimate the distribution´s parameters, an optimization algorithm is used as suggested in. An additional result is the derivation of the even moments of Ward´s KA distribution. This result is given in the appendix because it is needed for the comparison of the KA distributions.