Comprehensive Analysis, Processing And Interpretation Of The Full Tensor Gravity Gradient Data

Gravity field is an inherent physical property of earth, which can reflect thedensity distribution, the laws of motion and change of the earth material. Gravitymeasurement is the direct reflection of the gravity field variation. However, with thediversification of the measurements (ground, marine, airborne and satellite) and theimprovement of accuracy, we can measure the earth gravity or even the gravitygradient anomaly. Compared with traditional gravity measurement data, gravitygradient tensor data include more high frequency information, which can be used tostudy the earth’s interior structure, mineral resources distribution and otherinformation detailedly. In order to process and interpret the full tensor gravity gradientdata more accurately, we need to research the design principle of the full tensorgravity gradiometry, analyze the instrument noise, so that we can filter the noise of thegravity gradient data better to extract the real signals of geological bodies forinterpretation.At first, this paper use the theory model to analyze the advantages of gravitygradient measurement compared with gravity measurement, which provided the basisfor selecting the topic. The gravity gradient tensor data can either be measured orcalculated numerically from the gravity data. However, the calculated gravity gradienttensor data include the same information with the measured gravity data at most. Ingeneral, the calculated data will cause information loss, and do not add information.Therefore, we usually get the gravity gradient tensor data by measurement. Bycomparative analysis, the high frequency signal of gravity gradient anomaly canreflect the relatively short wavelength. However, the low frequency signal of gravityanomaly can reflect the relatively the long wavelength. Therefore, combining the highfrequency component of gravity gradient data and the low frequency component ofgravity data, we can get the enhanced gravity anomaly data, which retained all theinformation of gravity and gravity gradient data. The enhanced gravity anomalybroadens the frequency bandwidth scope compared with traditional gravity data. Thiswas accomplished by cosine square filter.At present, many research institutions are studying a variety of types of gravitygradiometers. However, only the gravity gradiometer based on the rotating accelerometer has been put into production. Therefore, for the high precision fulltensor gravity gradiometry used in airborne moving platform，this paper study thedifference composite structure of gradiometry，which have12accelerometers installedon three different rotation disc. Upon confirmation of the structure which can restrainthe common mode acceleration，reduce the interference of extent environment andhave the advantage of high precision detection，this paper analyze the error sourcesand influences in gradiometer measurements mainly. Research shows that maineffects include instrument inherent random noise and external deterministic noise. Inorder to describe the influence quantificationally，this paper derived the measuringequation of FTG in the dynamic environment；separated the three main intrinsicfactors: Accelerometer’s performance mismatch；Platform instability and Discrotating speed unstable and analyzed the noise level of intrinsic factors in timedomain and frequency domain. By using Simulink system，experiment can obtain thenoise level of intrinsic factors and put forward the inhibition scheme. Aiming at theexternal deterministic noise，we also analyzed the influence of changing of attitudeand quality in flight to measured gradients and put forward a method to correct theself-gradient based on a point mass. Every gravity gradient data corrected byintrinsic noise and self-gradient noise still include amount of random noise. And thenoise levels are different. This paper use the odd-even grid method to quantifyestimate the noise, and providing accord for noise filtering.Gravity gradient tensor data include higher frequency signals than gravity data,which can help to delineate small scale anomaly. However, measurements of fulltensor gradiometry are contaminated by high frequency random noise. Separation ofnoises from high frequency signals is one of the most challenge tasks in gravitygradient tensor data processing. Combining multiple gradient tensors measured by fulltensor gravity gradiometry simultaneously can effectively suppress the random noise.Based on the constraint that the real gradient signals meet the Laplace equation, wederived the series solution of the gravity gradient tensor data in Cartesian coordinatesystem, and use them to fit the measured gravity gradient data. The fitted results arereal gradient signals and the unfitted components are noise. In the processing of fitting,we use the optimal linear inversion method to solve the coefficient of the equation byintroducing the noise weight matrix and energy weight matrix, and then use the solvedcoefficient to get the real gravity gradient signals. After noise filtering, the full tensorgravity gradient data only include the real gravity information, which can be moreaccurate for interpretation. The interpretation of gravity gradient data often needs to obtain the horizontal position and the depth of the geological bodies. The horizontalposition is usually determined by the edge detection methods. The depth scopeparameters are calculated by depth calculation method.Edge detection plays an important role in the interpretation of gravity gradientdata, which has been widely used to delineate the edges of the sources. Sometraditional methods cannot display the edges of shallow and deep sourcessimultaneously. Although some method can identify the edges, the edges are notprecisely and some details are not displayed. Also, most of the methods are proposedfor gravity data. There are few methods are designed for gravity gradient tensor dataspecially. According to the characteristics of gravity gradient tensor data which canoutline more small geological bodies, this paper proposed the improved horizontalanalytic signal methods, enhanced horizontal directional total horizontal derivativesand improved structure tensor method to detect the edges of full tensor gravitygradient tensor data. In order to equalize the edge signal amplitude, we make somenormalization to the proposed methods. These methods have been demonstrated onthe synthetic model data and real measured data to show the advantages. Comparedwith some traditional edge filters, the new methods are more reliability and practical.The depth calculating of the causative bodies from the gravity and gravitygradient anomalies has been developed for a long time. There are many differentmethods for different type of sources. The rapid imaging method is a development hotspot in recent years, which can display the distribution of the underground geologicalbodies, and avoid the disadvantages of time-consuming and large memoryconsumption. This paper uses the Depth from Extreme Points (DEXP) to image theburied bodies, which can reflect the depth. However, the traditional DEXP methodneeds to know the structure index of the causative bodies. If we give a wrongstructure, the DEXP method will lead an error. Therefore, this paper uses the ratio ofdifferent order vertical derivatives of gravity anomaly to EDXP transform, which canimage the depth of geological bodies without the influence of structure index. Then,we can calculate the structure index with the imaged depth from the scaling function.This method has been tested on the model data and real measured Vinton Domegravity gradient data, which both get the accurate results.