S.E. Grigas - Ph.D. (Phys.-Math.), JSC «Corporation «Kometa», Moscow. E-mail: stanislav.grigas@gmail.com
D.Ts. Litovchenko - Dr.Sc. (Eng.), JSC «Corporation «Kometa», Moscow. E-mail: dc_litovchenko@mtu-net.ru
A.A. Skorynin - Ph.D. (Phys.-Math.), JSC «Corporation «Kometa», Moscow. E-mail: skoraleks@ya.ru
M.V. Chekmarev - Ph.D. (Eng.), JSC «Corporation «Kometa», Moscow. E-mail: chekmarev.mihail@gmail.com

Since the detection performance of synthetic aperture radar (SAR) is usually limited by radar returns from the underlying surface, radar clutter analysis plays an important role in SAR design. The first model for sea clutter was the Gaussian model which is valid for low resolution SAR images, where the number of scatters in each pixel is large. Analysis of high-resolution SAR images shows that sea clutter statistics deviates from Gaussian significantly. Thus the development of non-Gaussian sea clutter models is of great interest for target detection above the oceanic surface. X. Qin et al. proposed a generalized gamma distribution (GGD) for modeling sea clutter in high-resolution SAR images. In this paper we used GГD to calculate probabilities of detection for low observable aerial targets above the sea surface. The calculations were performed for small size Swerling I target, single-cell false alarm probability 1E-7, probability of detection from 0.05 to 0.95 and several sea clutter models: Gaussian, Log-Normal, Weibull and GGD. Signal-to-clutter ratio (SCR) obtained using GGD is 5-16 dB higher in comparison with other models. The reason is that GGD is the heavier-tailed than other distributions, thus in constant false alarm rate processing it leads to higher detection thresholds. To maintain the probability of detection at a desired level it requires to operate at higher SCR. Thus using GГD which can fit sea clutter in high-resolution images better than many other distributions allows us to obtain more accurate results for SAR detection performance.