The Study Of Ground Penetrating Radar Impedance Inversion Method And Its Application

Ground penetrating radar(GPR)detects and understands subsurface media by transmitting and receiving high-frequency electromagnetic(EM)waves.As instruments of GPR become more stable and reliable,quantitative amplitude information spawns another generation of data analysis and interpretation tools based on inversion method to image subsurface material properties.The parameterized result is more meaningful and effective than conventional amplitude profile.For on-ground GPR,current researches mainly include two schemes:velocity analysis and iterative inversion for permittivity.Velocity analysis provides a common way to estimate parameters but it gives limited resolution and certainties.Traditional permittivity inversion methods such as full-waveform inversion can return accurate estimates but the strong nonlinear make them unstable and still far from practical use.At the same time,most of them employ multi-offset GPR data,pretty rare compared to common offset data,whose inversion is rather promising in various applications.Common offset GPR data are considered as applicable to convolution model,based on which the impedance inversion is a generalized linear inversion strategy.Theoretically,GPR impedance inversion could obtain high resolution with low cost in calculation.The result in form of EM impedance could be transformed to other essential parameters for providing more geological information.Thus,it is of great value to study the impedance inversion method of GPR data and its application.In this paper,the concept of impedance,which is the most important physical parameter in this study,is expounded firstly.Both acoustic impedance and EM impedance are reviewed and summarized in detail,while the connections between them are also discussed.Next,the basic principle and parameters involved of GPR is elucidated,and survey mode,origin of noise and related data processing methods are briefly introduced around the purpose of EM impedance inversion.Then,principles of different techniques of GPR impedance inversion are elaborated together.On this basis,three techniques of GPR impedance inversion,that is,sparse spike deconvolution based on reflectivity model,band-limited impedance inversion based on deconvolution and broadband constrained inversion based on EM impedance model are studied respectively.For the sparse spike deconvolution,several solutions to extract wavelet especially deterministic wavelet are described,and two numerical tests are accomplished to verify deterministic wavelet sparse spike deconvolution's ability of identifying thin bed and resisting noise.This method also performs well in detecting seawalls.For the band-limited impedance inversion,a case study of Mudu city site in Suzhou,China is taken as an example of new workflows based on it.The result of the study shows that GPR impedance inversion is a feasible and effective method for imaging and characterization of buried archaeological remains.For the broadband constrained inversion,a numerical test based on a stochastic model is performed to exhibit its advantages over band-limited impedance inversion,which include that it's more reliable when constructing the initial model and there is no accumulation of errors in the process.This study improves the capability of investigating stochastic model with fluctuating interfaces.In general,the results show that the cost in human labor and calculation of GPR impedance inversion is low yet high resolution is provided.These methods could improve GPR's application level.Among them,band-limited impedance inversion is relatively fast,while it can also be performed with few well logs.The deconvolution method is very essential,and sparse spike deconvolution has many advantages over conventional deconvolution methods.Broadband constrained inversion provides more accurate and reliable results,while it's also demanding for it needs more prior information form well logs.Based on those impedance inversion techniques,new workflows can be put forward for different applications,which help to improve the detection level of GPR.The research result of this paper offers effective means for estimating subsurface properties,and is of important scientific and practical value.