The Study Of Growth Point Inversion Method Based On Gravity And Magnetic Data

Inversion is an important tool for quantitatively interpreting gravity and magnetic anomalies,which uses measured field data to calculate the properties,geometry,and location of geological bodies.Inversion methods can be classified into three categories:imaging methods,geometry inversion methods,and property inversion methods.In recent years,property inversion methods based on regularization theory have developed rapidly due to their many advantages,while geometry inversion methods have developed more slowly.This paper focuses on the growth point method and aims to promote the development of geometry inversion methods.The growth point method is based on the idea that any inversion method must add additional prior information to obtain the true distribution of underground sources(to reduce ambiguity).However,sometimes there is insufficient prior information available(unconstrained).Therefore,this paper proposes an improved geometry inversion method that can incorporate prior information into the inversion calculation under constrained conditions and estimate the anomalous body model using other data processing methods under unconstrained conditions.Furthermore,as a geometry inversion method,this method should have the advantages of high computational efficiency and clear boundary interpretation of the interpreted model.The growth point method is a systematic search method and an improvement on the existing two geometry inversion methods: the growing body method and the planting method.This paper first discusses the advantages and disadvantages of existing geometry inversion methods and then proposes an improved geometry inversion method based on the inversion idea of "growth," which is called the growth point method to distinguish it from existing methods.The growth point method retains the concept of "seed." In addition,to enable inversion to adapt to both constrained and unconstrained conditions,this paper adopts the approach of dividing the total misfit into data fitting and model fitting terms,following the regularization theory.The model fitting term restricts the growth space.The calculation of the seed and growth space is independent of the inversion and can be directly set by prior information or estimated by imaging and boundary recognition methods.In the case of prior information,the seed can be set and the vertical and horizontal boundaries of the growth space can be defined directly through geological surveys and logging data,which are used as a constraint in the inversion calculation.For the case of no prior information,this paper uses the correlation imaging method to calculate the initial seed position and uses the Tilt method to define the growth space.These two data processing methods are classical and are briefly introduced in this paper,and their influencing factors are analyzed.This improved geometry inversion method has the advantages of high boundary resolution and computational efficiency.Using the planting method for growth,i.e.,calculating only the adjacent unit cells to the existing model at each growth step,greatly improves the inversion efficiency.In a single model experiment,the computation time of the growth point method is only 1/6 of that of the growing body method.The model constraint term enables this inversion method to incorporate prior information constraints,while the combination with correlation imaging and Tilt method enables this method to perform inversion without any prior information.This feature meets our expectations for an inversion method that can adapt to both constrained and unconstrained conditions.However,this method still has some issues that need to be addressed,such as finding a more optimal replacement for the parameter "f" and selecting optimization methods with better performance and stronger ability to find the global optimal solution.