The Application Of Semiparametric Model In Airline Network Adjustment Of Airborne Gravimetry

Airborne gravimetry is a new gravity measurement technology, which has significant advantages compared with traditional gravity methods of measurement, it can map the gravity field of a big area rapidly and continuously for its increased speed, greater range, significantly low cost and the potential to reach anywhere. There are very broad application prospects for the earth sciences, geodesy and modern military. But airborne gravimetry is a dynamic measurement, measurement platform is in continuous high-speed state of motion, so the observed value will be influenced by various errors in the measurement process. Although it is possible to reduce the systemic errors by improving the instrument accuracy, by correcting formula and selecting appropriate digital filtering techniques to isolate the part of the interference with the signal. But for some low-frequency errors, it can not be eliminated by low-pass filter. So in actual measurement, the tracks of an airborne gravity survey are generally planed to join each other, discrepancies between measurements on different tracks may exist at the intersection because airborne gravity data are unavoidably affected by different kinds of error source, so we use the network adjustment to reduce systematic errors of airborne gravity under the condition of cross-points’difference. The traditional network adjustment methods are grafted from the network adjustment of the marine gravimetry, based on the discrepancies of different lines, we use time-related functions on behalf of the systematic errors, according to the principle of least squares to solve the model parameters of the measuring line, then compensating the systematic errors. Based on the imperfection of the traditional methods, in the paper, the author introduce of the semiparametric model in the network adjustment of airborne gravimetry, which can separate the gross errors from the systematic errors of the observed values. This new methods can provide more reliable results. This paper is organized as follows:In the first chapter, it mainly discusses the research status of the network adjustment and gross errors for airborne gravimetry. As the traditionalmethods of network adjustment in airborne gravimetry mainly was transplantated from marine gravity measurements, this paper presents a new method of network adjustment, applying semiparameter model to the network adjustment of airborne gravimetry, then it discusses the theoretical significance and practical value of this new way.In the second chapter, firstly it discusses the reduction of gravity value and the search of cross-points, that is because the network adjustment is based on the difference of cross-points and the measuring line is not at the same height.Then it mainly introduces four methods of traditional network adjustment for airborne gravimetry, such as adjustment of half-systematic error, adjustment of self-calibrating, adjustment of self-calibrating for significant lines and two-step processing. These traditional methods are transplanted from network adjustment of marine gravity survey. Adjustment of half-systematic error is a fixed value, besides that, other methods are based on the principle of least squares. Finally, through simulation examples, it verifies validity of the traditional methods, and sums up the advantages and disadvantages of various methods. For adjustment of half-systematic error is a fixed value which can not reflect the variation of the system errors. For adjustment of self-calibrating, by introducing the value of virtual observatory to solve the rank defect, getting weight by experience to calculate model parameters, so the precision depends on prior weight. For adjustment of self-calibrating for significant lines, it can solve the problem of rank defect, but it is not desirable because that it doesn’t compensate the non-significant lines. For two-step processing, it solves the problem of rank defect and does not consider the determining of prior weight, but its accuracy is over-reliance on the accuracy of the observed values of the crossing points.In the third chapter, based on the advantages and disadvantages of the traditional methods, the semiparametric model is introduced into the network adjustment of the airborne gravimetry. In the new model, non-parametric component standing for the model errors is introduced into the traditional adjustment model. In the chapter, it proposes two network adjustment models of airborne gravimetry semiparametric model of self-calibrating and semiparametric model of two-step processing. Then it deduces the scope of applicability of the semiparametric model, and it tests the hypotheses for systematic errors and gross errors using semiparametric model. Finally, using a simulation example (small amount of data) to verify the feasibility that introducing semiparametric model into the network adjustment of the airborne gravimetry, then using the simulated examples to verify that semiparametric model can detect and separate gross errors, and it excepts the new methods can be used into the network adjustment of the airborne gravimetry.In the forth chapter, in order to further verify the feasibility that semiparametric model can be applied to network adjustment of the airborne gravimetry and the extraction of gross errors, choosing datas from an actual survey area, using the traditional model and semiparametric model to compensate the systematic errors of the measured line, finally, using semiparametric model to extract gross errors.Chapter Five is the paper’s concludes. In this chapter, by appropriate choice of error’s model, the regularization matrix and smoothing factor, using semiparametric model for airborne gravity survey line network adjustment, separating system errors and gross errors, to overcome the deficiencies of traditional network adjustment of airborne gravimetry. But for the actual measurement of large-scale network, we also need to conduct further studies and research how to improve its reliability and speed.