It has been shown that for“high speed”sediment the effects of layered structure of the sedimentare not significant for a certain range of frequency at small grazing angle,and the reflection behaviorapproaches to the Rayleigh reflection specifled by the surficial sediment pararneters(p,c,a).In thatcase the frequency dependence of reflection loss is mainly due to scattering of the surface roughness.If the scattering loss is considered by using the Bass perturbation theory,the reflection loss at smallgrazing angle can be written as: -In|νφ={Q_0(p,c,a)+Q_1(f,σ)}φwhere Q_0 is the term due to Rayleigh loss independent of frequency,Q_1 is the term due to scatteringloss depending on frequency f and the spectral density of surface roughness.Some experiments,from which parameter Q can be extracted,are analysed.It is found,that in given frequency range(0.5—3.0kHz)the magnitude and the tendency of variation of the parameter Q extracted fromdiffernt experiments are distributed in the permissible region predicted It has been shown that for “high speed” sediment the effects of layered structure of the sedimentare not significant for a certain range of frequency at small grazing angle, and the reflection behaviorapproaches to the Rayleigh reflection speculated by the surficial sediment pararneters (p , c, a) .In thatcase the frequency dependence of reflection loss is mainly due to scattering of the surface roughness. If the scattering loss is considered by using the Bass perturbation theory, the reflection loss at small grazing angle can be written as: -In | νφ = {Q_0 (p, c, a) + Q_1 (f, σ)} φwhere Q_0 is the term due to Rayleigh loss independent of frequency, Q_1 is the term due to scatteringloss depending on frequency f and the spectral density of surface roughness.Some experiments, from which parameter Q can be extracted, are analyzed. It is found, that in given frequency range (0.5-3.0 kHz) the magnitude and the tendency of variation of the parameter Q extracted from doiffernt experiments are distributed in the permis sible region predicted