Study And Application Of High-precision Methods In Potential-field Data Processing

Potential-field data processing is an important part of potential-field data for geological-geophysicalinterpretation. with the imminence requirement for high resolution data and fine division of geological bodies,to study methods of high-precision in potential-field data processing has become an topic paid more attentionby geophysicists.Processed data of potential-field is mostly completed in wave number domain. However, the results ofsome calculations such as downward continuation, derivative calculation, reduction to the pole of magneticanomaly and so on, are unstable at times using directly from FFT as their transformation factors cansignificantly amplify some wave number signals. In order to solve this problem, geophysicists proposed aseries of methods to improve computational stability and raise the precision of calculation, and iterativemethod is attached widespread attention because of better results in applications than others.Based on the previous research works and the idea of iterative method, firstly, this paper presents Taylorseries iteration for downward continuation of potential-field to solute its intrinsic ill-posed problem. Theconvergence of this method is proved, the convergence speed of the term N=0and N=1is comparativeanalyzed. When Taylor series expansion term equals zero, Taylor series iteration is concordant with theintegral-iteration method in wave number domain. Model tests indicate that the iteration method of N=1hasvery fewer iteration times and better preserved amplitude than N=0when the depth of downward continuationis large and the continuation error of the term N=1equals N=0.Secondly, In order to improve calculation accuracy of some methods whose calculation is unstable inpotential-field, the paper gives a unified iterated mode called iterative filtering method. And what’s more, ascheme to resolve the problem of choosing the optimal number of iteration, is called the difference of twodifference cross-correlation coefficients. On this basis, I discuss the effect mechanism of calculation resultseffected by using different iterative times. According to the number of iteration increasing, the results can bedivided into three calculation zones including suppression zone, balance zone and interference zone. Modeltests of derivative calculation, downward continuation, and the reduction to the pole, are indicate that resultsof the iterative filtering presented in this paper has stronger stability and higher precision than other proposediterative methods. Meanwhile, the results further test and verify the validity of the method for selecting theoptimal number of iteration.And then, the iterative method is introduced into potential-field anomaly separation, normalized totalgradient of gravity, and inversion of density interface. Separating potential-field anomaly using the iterativefiltering, is a modified method in allusion to the local field which has a part of component of the regional field. The idea comes from iteration, but its physical mechanism is different from iterative filtering aboved. Modeltest of potential-field separation verifies the validity of using iterative filtering to separate potential-fieldanomaly. Normalized total gradient of gravity can be used to detect oil and gas reservoir. Using cut-offsmoothing method to calculate normalized total gradient of gravity is presented under enlightenment ofiterative method. This method can preserve more low frequency signals and suppress stronger high frequencyinterference than conventional methods, and model tests prove that cut-off smoothing has the advantages ofshorter trial computation and higher resolution than general methods. In order to avoid downwardcontinuation influence on density interface inversion, a new method called linear iterative inversion, has goodpoints about strong stability, high percison and low dependence to the number of iteration, proved by modeltests.Moreover, edge detection is also an important research content in this paper. Edge detection ofpotential-field is commonly used in recognizing edges of geological bodies. Potential-field anomalies have theinformation of field sources’ edges, but the information extraction relies on data processing. At present, thereare many methods can used to recognize edges, but almost of these existent methods have the disadvantagesthat the detected edges are blurred and the results are susceptible to interference. In this situation, The paperproposes normalized mean square error ratio, normalized differential, optimal auto-ratio of vertical gradient,and modified edge-detection methods.Normalized mean square error ratio based on the directionality of boundary anomaly and the datavolatility evaluated by mean square error, and it can display the edges of geological bodies on different deepsimultaneously.Normalized differential is according to the relationship between three-directional difference and thecharacter of potential-field anomaly in position of structured edges, and faulted structures with differentgrades can be recognized by using different radius and different order.The optimal auto-ratio of vertical gradient is used to clear interference and improve the ability of edgerecognition. Mathematical implication of the auto-ration is elaborated and physical meaning of this method isexplained. The optimal auto-ratio method can not only finely detect edges of field sources, but also candelineate different kinds of anomaly from geological bodies.The paper gives six modified edge-recognition methods including the total horizontal gradient, tilt angle,the total horizontal gradient of tilt angle, Theta map, normalized standard deviations, and normalized meansquare error ratio. Model tests indicate that modified methods can improve the ability of edge recognitionrelative to the original methods.At last, new methods presented in this paper and the conventional methods are applied to gravity andmagnetic anomalies of Yalujiang basin whose geological conditions are very complex, the data processingresults further confirmed that new methods can improve the precision of data processing. And according to theabove results, geological data, and electric data,32faulted structures including12faults can proved by three two-dimensional resistivity inversion section is detected and the basin is divided into four first-order tectonicunits and11second-order tectonic units including2concealed depressions.