Geoacoustic Parameter Inversion Based On Backscattering Strength

Seafloor plays an important role in acoustic propagation as the boundary of ocean waveguide.The acoustic properties of seafloor are unavoidable in reliable acoustic propagation models and source localization by matched-field processing.However,it is usually difficult to measure the seafloor parameters in large range,even at much cost.Recently,the inversion method employing acoustic field data to estimate seafloor parameters has been an effective manner to realize remote sensing,which possesses a broad application prospect.Acoustic scattering is the result of interaction between sound and seafloor,which certainly contains plenty of information associated with the seafloor.Consequently,inversion method based on backscattering strength to estimate seafloor parameters is investigated and verified in this paper,where the forward model of backscattering strength is established and backscattering strength measurement of sandy sediments at the bottom of a water tank is conducted,under the laboratory environment.Comparing with traditional matched-field inversion based on acoustic propagation model,this method has many advantages involved in simple algorithm,easy realization and available more comprehensive seafloor information.In the forward model,the seafloor is viewed as a two-phase system consists of solid frame and pore water,which is described by effective density fluid model(EDFM,sometimes also known as effective density fluid approximation)applying to softy sediments,such as mud(including clay and silt)and fine sand,as a simplification of Biot theory.In EDFM,the effective density is a complex number related to frequency.Both scatterings arising from roughness of seafloor surface and volume inhomogeneity of sediments are considered,which are described by roughness scattering model of composite roughness and volume scattering model of small-perturbation fluid approximation,respectively.Take the dispersion property of sediments into account,where the sound-speed dispersion and frequency dependence of attenuation are calculated according to physical parameters out of inversion and EDFM.In summary,the forward model requires 10 input parameters and sound speed and attenuation are indirect inversion parameters in the form of speed ratio of sediments and sea water and loss parameter.Object function used to scale the mismatch between measured data and prediction value of backscattering strength is established on the assumption that the data error subjects Gaussian distribution.On the one hand,to get the optimization solution,a two-step hybrid optimization algorithm combining differential evolution algorithm and particle swarm optimization is proposed,which is used to overcome disadvantages of single algorithm and effectively improve the accuracy of inversion results.On the other hand,to analyze the uncertainty of inversion parameters and correlation between parameters,Bayesian inversion is adopted to present marginal probability distribution and covariance matrix of inversion parameters through the posterior probability density.The numerical simulation shows that the indirect inversion method performs strong robustness in inversion of geoacoustic parameters,including sound speed and attenuation.The estimation of loss parameter is satisfied in particular.Finally,the efficiency of the inversion method is verified through comparison between inversion results of the sandy sediments at the bottom of the water tank with measured parameters of collecting sand sample.The measured parameters include porosity,mean grain size,mass density of grains,sound speed and attenuation in high-frequency band.In measurement of sound speed and attenuation of the sand sample(in the frequency range of 90kHz-170kHz),a wide-band measurement based on pulse compression technique is proposed,by which the sound-speed dispersion and frequency dependence of attenuation in measured frequency band can be obtained continuously.Comparing with the narrow-band measurement,it is found that the wide-band measurement of sound speed and attenuation exhibits certain fluctuation,especially for attenuation at the second half of measured frequency band,where the bias is very large.This phenomenon probably arises from lower time-delay resolution of the wide-band signal in measurement,and lead to partial overlapping between correlation peak of directive wave and that of vicinity multi-path.Comparing with the simulation results,the uncertainty of inversion parameters derived from measured data of backscattering strength somewhat increase and the correlation between partial parameters further decrease,which is the result of larger data error than the case of simulation.Comparing with the measured value of parameters,the inversion result of porosity is a little smaller,while the sound speed is a little larger in high-frequency band.Besides the influence of data error,the deposition state has been changed in the process of collection,transition and measurement,which can also lead to certain contrast of physical and geoacoustic parameters between the measured value and that of true value.