alexa
Reach Us +44-1904-929220
Gravity Anomaly Data Reconstruction Based on Improved Interior Point Method in Airborne Gravimetry | OMICS International
ISSN: 2165-784X
Journal of Civil & Environmental Engineering
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on Medical, Pharma, Engineering, Science, Technology and Business

Gravity Anomaly Data Reconstruction Based on Improved Interior Point Method in Airborne Gravimetry

Yapeng Yang* and Meiping Wu
Department of Automatic Control, College of Mechatronics and Automation, National University of Defense Technology, Deya Street 109, Changsha 410073, China
Corresponding Author : Yapeng Yang
Department of Automatic Control
College of Mechatronics and Automation
National University of Defense Technology
Deya Street 109, Changsha 410073, China
Tel: +86-137-8618-8163
Fax: +86-731-8457-6305 (ext. 8212)
E-mail: [email protected]
Received February 10, 2015; Accepted March 24, 2015; Published April 04, 2015
Citation: Yang Y, Wu M (2015) Gravity Anomaly Data Reconstruction Based on Improved Interior Point Method in Airborne Gravimetry. J Civil Environ Eng 5:170. doi:10.4172/2165-784X.1000170
Copyright: © 2015 Yang Y, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Visit for more related articles at Journal of Civil & Environmental Engineering

Abstract

Strap-down air borne gravimeter is the core sensor of strap-down gravimetry system. In this paper, A new strap-down airborne gravimeter called SGA-WZ has been developed by the Laboratory of Inertial Technology of the National University of Defense Technology. Precise gravity anomaly data reconstruction is the key technology in airborne gravimetry based on the SGA-WZ. This paper explores a method for large scale precise gravity anomaly data reconstruction using the theory of Compressed Sensing (CS). Based on the CS theory, the gravity anomaly data reconstruction can be transformed into the one-norm convex quadratic program to be solved by several methods such as interior point methods (IPM). This paper presents an improved IPM for performing the large scale gravity anomaly data reconstruction precisely that uses the Preconditioned Conjugate Gradients Algorithm (PCG) to compute the search direction. And then, a flight test has been carried out in China with SGA-WZ. The test results have shown that the PCG-IPM algorithm can reconstruct the large scale gravity anomaly data with higher accuracy than the existing nearest interpolation method in airborne gravimetry.

Keywords
Airborne gravimetry; Gravity anomaly data reconstruction; Compressed sensing; Improved interior point method
Introduction
The gravity is one of the primary forces on the earth, which affects all the physical events of the earth and it’s near space. Gravity survey is significant in many fields, such as geology, seismology, geophysics, navigation technology, military science, and resource investigation, etc. Airborne gravimetry is an important measuring technology of acquiring the earth gravity. Compared to traditional terrestrial gravimetry and shipborne gravimetry, airborne gravimetry uses the measuring instruments taken on an aircraft, so which can get extensively and precisely well-distributed information of the earth gravity field economically and rapidly [1]. Meanwhile, some special terrains are difficult to access by terrestrial gravimetry and shipborne gravimetry, such as desert, marsh, glacier, mountain and forest, but airborne gravimetry can measure gravity in these areas effectively [2]. Just because of this, airborne gravimetry has become one of the hot research fields in geophysics.
Gravity anomaly data reconstruction is an important research topic in airborne gravimetry. Nyquist and Shannon showed that signals can be exactly recovered from a set of uniformly spaced samples taken at the Nyquist rate of twice the highest frequency present in the signal of interest [3,4]. Essentially, airborne gravimetry is a sub-Nyquist sampling method, and then, the precise gravity anomaly data reconstruction becomes a difficult work. In 2006, Candès, Romberg, Tao and Donoho presented a new theory named as CS, which offers the opportunity to perform the gravity anomaly data reconstruction precisely. The theory showed that a signal having a sparse representation can be reconstructed exactly from a small set of linear measurements [5-8]. Taking into account the sparsity of airborne gravimetry by the Discrete Fourier Transform (DFT), the gravity anomaly data reconstruction can be transformed into the one-norm convex quadratic program, which can be solved by standard convex optimization methods such as IPM algorithms. The recently developed computational methods for one-norm convex quadratic program include path-following methods [9,10], bound optimization methods, iterated shrinkage methods [11], and gradient projection algorithms. This paper mainly describes the PCG-IPM algorithm for performing the large scale gravity anomaly data reconstruction precisely. The method uses the PCG algorithm to compute the search direction [12].
Materials and Methods
Compressed sensing is a rapidly growing field that has attracted considerable attention in geophysics, seismology, applied mathematics, and computer science. Especially, reducing the sampling rate and increasing resolution in airborne gravimetry can reduce economic cost, improve survey efficiency, and improve data quality. Let image be an unknown original gravity anomaly, image is the sampling gravity anomaly, and image is the compressed sensing matrix. If z is sparse in a transform matrix image and the vectors ΙΈi are well chosen, the original gravity anomaly can be recovered from a amazingly small number of linear measurements m. And then, based on the theory of CS, the gravity anomaly data reconstruction can be studied by solving a problem of the form:
image    (1)
Where λ > 0 is the regularization parameter, x is the sparse coefficient matrix, and image is the one-norm of x. When Ψ is invertible, and let image, , Equation (1) can be rewritten as:
image (2)
This is the one-norm regularized least squares problem. By associating the linear inequality constraints, Equation (2) can be transformed to the convex quadratic program:
image (3)
The convex quadratic program can be solved by the PCG-IPM algorithm. This paper mainly studies the application of PCG-IPM algorithm in precise gravity anomaly data reconstruction. The method uses the PCG method to compute the search step, therefore, which can reduce the computation time for large scale problems with modest accuracy. Compared with the IPM that uses conjugate gradient methods to compute the search step, the PCG-IPM algorithm is able to solve the gravity anomaly data reconstruction with higher accuracy and smaller computational cost.
By associating the dual variable image, the Lagrange dual of Equation (2) can be formulated as:
image(4)
This dual problem is a convex optimization problem. Let v* is the solution of Equation (4), then we can say that v* is dual feasible [13]. Meanwhile, if P* is the optimal value of Equation (2), the dual feasible point v* supplies a lower bound of P*, i.e., image. For an arbitrary x of the one-norm regularized least squares problem, an simply computed bound of the sub-optimality x* can be obtained by constructing a dual feasible point:
image(5)
Because the point v* is dual feasible, G(ν*) is a lower bound of P*. The difference between G(ν*) and P* is named the duality gap denoted with η :
image(6)
where η ≥ 0 . When x* is the optimal solution of Equation (2), the duality gapη = 0 .
For the bound constraints image in Equation (3), we can define the logarithmic barrier:
image(7)
And then, Equation (3) can be reformulated to the convex function as:
image(8)
 
As the parameter t varies from 0 to ∞, we can obtain a series of the unique solution image of Equation (8). With each point image , we can construct the dual feasible variable v* shown in Equation (5). And then, the series of the point image leads to an optimal solution of Equation (2).
Generally, the solution of Equation (8) can be obtained with the Newton’s method:
image(9)
The solution of Equation (9) exactly is the search direction (Δx, Δu) . For the gravity anomaly data reconstruction, the PCG algorithm computes the search direction as an approximate solution of Equation (9) [14-16], so it gives a good tradeoff of computational complexity versus the data recovery precision.
In Equation (9), image can be formulated as:
image(10)
image(11)
With the PCG algorithm, the search direction (Δx, Δu) can be computed approximately [17-19], with a preconditioner:
image(12)
where image is symmetric and positive definite. We can see that Equation (12) is the approximation of Equation (10), so the cost of computing the search direction can be reduced.
After obtaining the search direction (Δx, Δu) by using the PCG algorithm, the new point is (x,u) + s(Δx,Δu) , where s>0 is the step size, which can be taken with the backtracking line search method as:
image(13)
If r ≥ 0 is the smallest integer that satisfies Equation (13), the step size is s=βr. Meanwhile, the stopping criterion of the PCG-IPM algorithm can be expressed as:
image(14)
Where image is the relative tolerance parameter for the PCG-IPM algorithm.
In conclusion, the procedure of PCG-IMP can be carried out by means of the basic steps as follows:
Step 1: Initialize the algorithm parameters: t : =1 λ , x : = 0 , image
Step 2: Compute the approximate solution of Equation (9) to obtain the search direction (Δx, Δu) , by using the PCG algorithm.
Step 3: Evaluate the step size s with the backtracking line search method, and update the new point (x,u) + s(Δx,Δu) .
Step 4: Construct the dual feasible variable v* shown in Equation (5), and compute the dual gap η shown in Equation (6).
Step 5: Update the parameter t.
Step 6: If the stopping criterion shown in Equation (14) is met, the PCG-IPM algorithm will come to the end. Else, return to Step 2.
Step 7: Obtain the estimate for the unknown gravity anomaly: image
Test Results and Discussions
From April 2010 to May 2010, the flight test was carried out in China. Among the eight flights in the whole flight test, the first and second flights were repeated lines. The other six flights were grid flights consisting of three flights of survey lines and three flights of control lines. In the test, we used the strap-down airborne gravimetry system, the core of which is strap-down airborne gravimeter. Because the strapdown airborne gravimeter has so many advantages of high reliability, low cost, unattended operation, and so on, it attracts broad attention of the researchers and develops rapidly [20]. In this paper, the Laboratory of Inertial Technology of the National University of Defense Technology has developed a strap-down airborne scalar gravimeter independently, as Figure 1 shows, which is the first home-grown system in China called SGA-WZ. In this paper, the SGA-WZ can be divided into two main components: the strap-down inertial navigation system [21,22] and the triad of accelerometers (Figure 2). The strap-down inertial navigation system is used mainly for measuring the specific force and the attiude of the aircraft. The accelerometers have a stability of ± 0.2 mGal, scale factor uncertainty of ± 3 ppm and random noise of ± 5 mGal. The gyroscopes have a stability of ± 0.004°/h and random noise of ± 0.002°/√h .To evaluate the PCG-IPM algorithm performance and the gravity anomaly data reconstruction precision, this study defines the performance indices with the standard deviation and SNR. SNR can be calculated by:
image(15)
Due to space limitations, this paper mainly analyzes the first flight denoted by F401, which consists of 6 lines. In this study, the algorithm parameters of PCG-IPM were set as:
image(16)
And the compressed measurement number was set as m=1618.
Figure 3 shows the reconstruction results of line 1 in F401 by the PCG-IPM algorithm. The comparisons between the original gravity anomaly and reconstruction gravity anomaly of line 1 in F401 are shown in Figure 3a. In this subgraph, the red dashed denotes the reconstruction gravity anomaly, the blue solid denotes the original gravity anomaly, and the two curves almost overlap each other if it doesn’t take the boundary effect into account. The reconstruction error of line 1 in F401 is shown in Figure 3b. Without regard to the boundary effect, the maximum reconstruction error of line 1 in F401 is 0.108 mGal, the minimum reconstruction error of line 1 in F401 is -0.132 mGal, and the standard deviation of reconstruction error of line 1 in F401 is 0.023 mGal. In practice, if the boundary data can be abandoned in the post-mission analysis, the boundary effect can also be ignored.As a consequence, Figure 3 demonstrates that the reconstruction results from the PCG-IPM algorithm approximate the original gravity anomaly very well and the reconstruction error is also very small. As mentioned above, the compressed measurement number was set as m =1618 , i.e., the compressed sampling rate was approximately 52%. These results convincingly show that the PCGIPM algorithm can reconstruct the gravity anomaly data with higher accuracy and fewer measurements. To contrast the gravity anomaly data reconstruction precision of the PCG-IPM algorithm, the paper also made the gravity anomaly data reconstruction with the existing Nearest Interpolation Method (NIP), which is one of the commonly used methodologies for data reconstruction in airborne gravimetry. Figure 4 shows the reconstruction results of line 1 in F401 by the NIP algorithm. The comparisons between the original gravity anomaly and reconstruction gravity anomaly of line 1 in F401 are shown in Figure 4a. Be similar to Figure 3a, the red dashed denotes the reconstruction gravity anomaly, and the blue solid denotes the original gravity anomaly. The reconstruction error of line 1 in F401 is shown in Figure 4b. The maximum reconstruction error of line 1 in F401 is 0.289 mGal, the minimum reconstruction error of line 1 in F401 is -0.293 mGal, and the standard deviation of reconstruction error of line 1 in F401 is 0.11 mGal. As a consequence, Figures 4 demonstrates that there is not the boundary effect existing in the reconstruction results of the NIP algorithm, but the gravity anomaly data reconstruction precision of the NIP algorithm is inferior to the corresponding one of the PCG-IPM algorithm. The comparisons of the algorithm performance between the PCG-IPM and NIP algorithm for F401 are shown in Table 1. In Table 1, taking an average, the standard deviation of reconstruction error and SNR of the PCG-IPM algorithm in F401 are 0.024 mGal and 58.277 dB, respectively. Correspondingly, the standard deviation of reconstruction error and SNR of the NIP algorithm are 0.109 mGal and 45.283 dB, respectively.
Conclusion
Strap-down airborne gravimeter is the core sensor of strap-down gravimetry system. This paper explores a method for large scale precise gravity anomaly data reconstruction based on the strap-down airborne gravimeter. Precise gravity anomaly data reconstruction is the key technology in airborne gravimetry. In this paper, our work can be summarized as follows:
(1) The strap-down airborne gravimeter (SGA-WZ) mentioned in this paper is the first system with this type in China. The system is an inexpensive, robust, flexible tool to carry out airborne gravimetry.
(2) Based on the theory of CS, the paper presents the PCG-IPM algorithm applied to perform the large scale gravity anomaly data reconstruction precisely.
(3) A flight test was carried out in China. By the comparison between the PCG-IPM algorithm and the NIP algorithm, the each performance index from the PCG-IPM algorithm is superior to the corresponding one from the NIP algorithm. These results convincingly show that the PCG-IPM algorithm can reconstruct the gravity anomaly data with higher accuracy and fewer measurements.
However, the boundary effect exists in the result of gravity anomaly data reconstruction with the PCG-IPM algorithm. Although we can abandon the boundary data, that will result in the waste of data and increase the survey costs. This problem should be solved in the future researchers.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgment
This work was supported by the National High Technology Research and Development Program of China (SS2013AA060402). The authors would like to thank the anonymous reviewers for their helpful comments which have improved the quality of the paper.
References

Tables and Figures at a glance

Table icon
Table 1

Figures at a glance

Figure Figure Figure Figure
Figure 1 Figure 2 Figure 3 Figure 4
Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Article Usage

  • Total views: 12251
  • [From(publication date):
    April-2015 - Feb 21, 2020]
  • Breakdown by view type
  • HTML page views : 8426
  • PDF downloads : 3825
Top