The masking background of the image produced by a sea bottom targets detection sonar, which is mainly composed of bottom reverberation, has been modeled after linear reception by a non-stationary Weibull distribution, the parameters of which depend on grazing angle and sea bottom type. Detection curves of a steady target are then presented for various shape parameters p of the Weibull interference (

) and various probabilities of false alarm. For a same (P
D, P
FA), these curves show that for p<2, it is necessary to have a higher signal-to-interference ratio than for p = 2 (Rayleigh distribution), whereas for p>2, a Lower signal-to-interference ratio is sufficient.