Radar sea‐clutter modelling using fractional generalised Pareto distribution

Abstract

A model for radar sea-clutter derived from the fractional differentiation of compound Gaussian distribution with inverse gamma texture is presented. Different stages of fractional calculus are conducted to obtain the new probability density function (PDF) and cumulative distribution function with fractional order. To investigate the fitting of the proposed statistical model to real data, the curve fitting estimation method based on the Nelder–Mead algorithm is used to optimise the shape parameter, the scale parameter and the fractional derivative order. Compared to the generalised Pareto model with the maximum likelihood estimator, it is shown that the proposed fractional PDF models well the tail of IPIX real data especially when medium range resolutions are at hand.