Magnetization clustering inversion — Part 1: Building an automated numerical optimization algorithm

Geophysics ◽  
2018 ◽  
Vol 83 (5) ◽  
pp. J61-J73 ◽  
Author(s):  
Jiajia Sun ◽  
Yaoguo Li
Keyword(s):  
Field Data ◽  
Learning Algorithm ◽  
Search Algorithm ◽  
Magnetic Data ◽  
Optimization Strategy ◽  
Inversion Algorithm ◽  
Data Set ◽  
Regularized Inversion ◽  
Automatic Search ◽  
Parameter Values

Magnetic data are among the most widely used geoscientific data for studying the earth interior in the oil and mining industries. However, interpreting magnetic data has been traditionally challenged by the presence of remanence. Recently, a new inversion algorithm, called magnetization clustering inversion (MCI), was developed by combining the classical Tikhonov regularized inversion with fuzzy [Formula: see text]-means clustering, an unsupervised machine-learning algorithm. This method has proven to be an effective tool for interpreting magnetic data complicated by remanence through synthetic and field data tests. However, the MCI algorithm in previous work requires users to specify the values of the weighting parameters for both smoothness regularization and clustering terms in the objective function. In practice, this entails many iterations of trial and error, and it consequently hinders the effective use of this inversion algorithm for iterative hypothesis testing and timely decision making. We have developed an automated strategy for determining the two weighting parameter values. Our algorithm of automatic search for optimal weighting parameters is based on an understanding of their roles and the complex interplay between them during an inversion. Our search algorithm works by alternately searching for one weighting parameter, whereas the other is fixed. A series of synthetic examples confirms the effectiveness of this automated optimization strategy. We also applied the automated inversion algorithm to a field data set from the Carajás Mineral Province in Brazil. The recovered magnetic anomalous features are highly consistent with known geology.

Separation of regional and residual magnetic field data

Geophysics ◽  
1998 ◽  
Vol 63 (2) ◽  
pp. 431-439 ◽  
Author(s):  
Yaoguo Li ◽  
Douglas W. Oldenburg
Keyword(s):  
Magnetic Field ◽  
Field Data ◽  
Power Spectra ◽  
Magnetic Data ◽  
Inversion Algorithm ◽  
Large Area ◽  
Data Set ◽  
Regional Sources ◽  
Magnetic Inversion ◽  
Susceptibility Model

We present a method for separating regional and residual magnetic fields using a 3-D magnetic inversion algorithm. The separation is achieved by inverting the observed magnetic data from a large area to construct a regional susceptibility distribution. The magnetic field produced by the regional susceptibility model is then used as the regional field, and the residual data are obtained by simple subtraction. The advantages of this method of separation are that it introduces little distortion to the shape of the extracted anomaly and that it is not affected significantly by factors such as topography and the overlap of power spectra of regional and residual fields. The proposed method is tested using a synthetic example having varying relative positions between the local and regional sources and then using a field data set from Australia. Results show that the residual field extracted using this method enables good recovery of target susceptibility distribution from inversions.

Mitigating remanent magnetization effects in magnetic data using the normalized source strength

Geophysics ◽  
2013 ◽  
Vol 78 (3) ◽  
pp. J25-J32 ◽  
Author(s):  
Mark Pilkington ◽  
Majid Beiki
Keyword(s):  
Magnetic Field ◽  
Remanent Magnetization ◽  
Field Data ◽  
Magnetic Data ◽  
Magnetization Distribution ◽  
Source Strength ◽  
Inversion Algorithm ◽  
Data Set ◽  
Magnetic Field Data ◽  
The Magnetic Field

We have developed an approach for the interpretation of magnetic field data that can be used when measured anomalies are affected by significant remanent magnetization components. The method deals with remanent effects by using the normalized source strength (NSS), a quantity calculated from the eigenvectors of the magnetic gradient tensor. The NSS is minimally affected by the direction of remanent magnetization present and compares well with other transformations of the magnetic field that are used for the same purpose. It therefore offers a way of inverting magnetic data containing the effects of remanent magnetizations, particularly when these are unknown and are possibly varying within a given data set. We use a standard 3D inversion algorithm to invert NSS data from an area where varying remanence directions are apparent, resulting in a more reliable image of the subsurface magnetization distribution than possible using the observed magnetic field data directly.

Adaptive sampling of potential-field data: A direct approach to compressive inversion

Geophysics ◽  
2014 ◽  
Vol 79 (1) ◽  
pp. IM1-IM9 ◽  
Author(s):  
Nathan Leon Foks ◽  
Richard Krahenbuhl ◽  
Yaoguo Li
Keyword(s):  
Potential Field ◽  
Adaptive Sampling ◽  
Field Data ◽  
Computational Cost ◽  
Large Field ◽  
Magnetic Data ◽  
Data Sampling ◽  
Data Types ◽  
Data Set ◽  
Potential Field Data

Compressive inversion uses computational algorithms that decrease the time and storage needs of a traditional inverse problem. Most compression approaches focus on the model domain, and very few, other than traditional downsampling focus on the data domain for potential-field applications. To further the compression in the data domain, a direct and practical approach to the adaptive downsampling of potential-field data for large inversion problems has been developed. The approach is formulated to significantly reduce the quantity of data in relatively smooth or quiet regions of the data set, while preserving the signal anomalies that contain the relevant target information. Two major benefits arise from this form of compressive inversion. First, because the approach compresses the problem in the data domain, it can be applied immediately without the addition of, or modification to, existing inversion software. Second, as most industry software use some form of model or sensitivity compression, the addition of this adaptive data sampling creates a complete compressive inversion methodology whereby the reduction of computational cost is achieved simultaneously in the model and data domains. We applied the method to a synthetic magnetic data set and two large field magnetic data sets; however, the method is also applicable to other data types. Our results showed that the relevant model information is maintained after inversion despite using 1%–5% of the data.

Computer Application in Geosciences: Imaging of the Subsurface by Structural Coupled Inversion of Potential Field Data

2014 ◽  
Vol 644-650 ◽  
pp. 2670-2673
Author(s):  
Jun Wang ◽  
Xiao Hong Meng ◽  
Fang Li ◽  
Jun Jie Zhou
Keyword(s):  
Potential Field ◽  
Field Data ◽  
Gravity Data ◽  
Computer Application ◽  
Magnetic Data ◽  
Imaging Method ◽  
Inversion Algorithm ◽  
Constant Property ◽  
Near Surface ◽  
Potential Field Data

With the continuing growth in influence of near surface geophysics, the research of the subsurface structure is of great significance. Geophysical imaging is one of the efficient computer tools that can be applied. This paper utilize the inversion of potential field data to do the subsurface imaging. Here, gravity data and magnetic data are inverted together with structural coupled inversion algorithm. The subspace (model space) is divided into a set of rectangular cells by an orthogonal 2D mesh and assume a constant property (density and magnetic susceptibility) value within each cell. The inversion matrix equation is solved as an unconstrained optimization problem with conjugate gradient method (CG). This imaging method is applied to synthetic data for typical models of gravity and magnetic anomalies and is tested on field data.

scholarly journals A fast methodology for large-scale focusing inversion of gravity and magnetic data using the structured model matrix and the 2-D fast Fourier transform

2020 ◽  
Vol 223 (2) ◽  
pp. 1378-1397
Author(s):  
Rosemary A Renaut ◽  
Jarom D Hogue ◽  
Saeed Vatankhah ◽  
Shuang Liu
Keyword(s):  
Fourier Transform ◽  
Fast Fourier Transform ◽  
Linear Systems ◽  
Large Scale ◽  
Surface Measurement ◽  
Magnetic Data ◽  
Uniform Grid ◽  
Data Sets ◽  
Inversion Algorithm ◽  
Data Set

SUMMARY We discuss the focusing inversion of potential field data for the recovery of sparse subsurface structures from surface measurement data on a uniform grid. For the uniform grid, the model sensitivity matrices have a block Toeplitz Toeplitz block structure for each block of columns related to a fixed depth layer of the subsurface. Then, all forward operations with the sensitivity matrix, or its transpose, are performed using the 2-D fast Fourier transform. Simulations are provided to show that the implementation of the focusing inversion algorithm using the fast Fourier transform is efficient, and that the algorithm can be realized on standard desktop computers with sufficient memory for storage of volumes up to size n ≈ 106. The linear systems of equations arising in the focusing inversion algorithm are solved using either Golub–Kahan bidiagonalization or randomized singular value decomposition algorithms. These two algorithms are contrasted for their efficiency when used to solve large-scale problems with respect to the sizes of the projected subspaces adopted for the solutions of the linear systems. The results confirm earlier studies that the randomized algorithms are to be preferred for the inversion of gravity data, and for data sets of size m it is sufficient to use projected spaces of size approximately m/8. For the inversion of magnetic data sets, we show that it is more efficient to use the Golub–Kahan bidiagonalization, and that it is again sufficient to use projected spaces of size approximately m/8. Simulations support the presented conclusions and are verified for the inversion of a magnetic data set obtained over the Wuskwatim Lake region in Manitoba, Canada.

2.5‐D inversion of frequency‐domain electromagnetic data generated by a grounded‐wire source

Geophysics ◽  
2002 ◽  
Vol 67 (6) ◽  
pp. 1753-1768 ◽  
Author(s):  
Yuji Mitsuhata ◽  
Toshihiro Uchida ◽  
Hiroshi Amano
Keyword(s):  
Frequency Domain ◽  
Regularization Parameter ◽  
Synthetic Data ◽  
Magnetic Data ◽  
Statistical Criterion ◽  
Model Parameters ◽  
Inversion Algorithm ◽  
Gas Field ◽  
Data Set ◽  
Transient Electromagnetic

Interpretation of controlled‐source electromagnetic (CSEM) data is usually based on 1‐D inversions, whereas data of direct current (dc) resistivity and magnetotelluric (MT) measurements are commonly interpreted by 2‐D inversions. We have developed an algorithm to invert frequency‐Domain vertical magnetic data generated by a grounded‐wire source for a 2‐D model of the earth—a so‐called 2.5‐D inversion. To stabilize the inversion, we adopt a smoothness constraint for the model parameters and adjust the regularization parameter objectively using a statistical criterion. A test using synthetic data from a realistic model reveals the insufficiency of only one source to recover an acceptable result. In contrast, the joint use of data generated by a left‐side source and a right‐side source dramatically improves the inversion result. We applied our inversion algorithm to a field data set, which was transformed from long‐offset transient electromagnetic (LOTEM) data acquired in a Japanese oil and gas field. As demonstrated by the synthetic data set, the inversion of the joint data set automatically converged and provided a better resultant model than that of the data generated by each source. In addition, our 2.5‐D inversion accounted for the reversals in the LOTEM measurements, which is impossible using 1‐D inversions. The shallow parts (above about 1 km depth) of the final model obtained by our 2.5‐D inversion agree well with those of a 2‐D inversion of MT data.

Structurally constrained impedance inversion

2016 ◽  
Vol 4 (4) ◽  
pp. T577-T589 ◽  
Author(s):  
Haitham Hamid ◽  
Adam Pidlisecky
Keyword(s):  
Objective Function ◽  
Seismic Data ◽  
Field Data ◽  
A Priori ◽  
Structural Constraints ◽  
Inversion Algorithm ◽  
Data Set ◽  
Impedance Inversion ◽  
Single Trace ◽  
Complex Geology

In complex geology, the presence of highly dipping structures can complicate impedance inversion. We have developed a structurally constrained inversion in which a computationally well-behaved objective function is minimized subject to structural constraints. This approach allows the objective function to incorporate structural orientation in the form of dips into our inversion algorithm. Our method involves a multitrace impedance inversion and a rotation of an orthogonal system of derivative operators. Local dips used to constrain the derivative operators were estimated from migrated seismic data. In addition to imposing structural constraints on the inversion model, this algorithm allows for the inclusion of a priori knowledge from boreholes. We investigated this algorithm on a complex synthetic 2D model as well as a seismic field data set. We compared the result obtained with this approach with the results from single trace-based inversion and laterally constrained inversion. The inversion carried out using dip information produces a model that has higher resolution that is more geologically realistic compared with other methods.

Joint inversion of surface and three‐component borehole magnetic data

Geophysics ◽  
2000 ◽  
Vol 65 (2) ◽  
pp. 540-552 ◽  
Author(s):  
Yaoguo Li ◽  
Douglas W. Oldenburg
Keyword(s):  
Field Data ◽  
Joint Inversion ◽  
Magnetic Data ◽  
Data Sets ◽  
Source Depth ◽  
Inversion Algorithm ◽  
Borehole Data ◽  
Weighting Functions ◽  
Surface Data ◽  
Geometric Decay

The inversion of magnetic data is inherently nonunique with respect to the distance between the source and observation locations. This manifests itself as an ambiguity in the source depth when surface data are inverted and as an ambiguity in the distance between the source and boreholes if borehole data are inverted. Joint inversion of surface and borehole data can help to reduce this nonuniqueness. To achieve this, we develop an algorithm for inverting data sets that have arbitrary observation locations in boreholes and above the surface. The algorithm depends upon weighting functions that counteract the geometric decay of magnetic kernels with distance from the observer. We apply these weighting functions to the inversion of three‐component magnetic data collected in boreholes and then to the joint inversion of surface and borehole data. Both synthetic and field data sets are used to illustrate the new inversion algorithm. When borehole data are inverted directly, three‐component data are far more useful in constructing good susceptibility models than are single‐component data. However, either can be used effectively in a joint inversion with surface data to produce models that are superior to those obtained by inversion of surface data alone.

Subsalt imaging by target-oriented inversion-based imaging:A 3D field-data example

Geophysics ◽  
2013 ◽  
Vol 78 (2) ◽  
pp. S47-S58 ◽  
Author(s):  
Yaxun Tang ◽  
Biondo Biondi
Keyword(s):  
Field Data ◽  
Imaging System ◽  
Normal Operator ◽  
Data Set ◽  
Image Domain ◽  
Depth Migration ◽  
Regularized Inversion ◽  
Computationally Intensive ◽  
Subsalt Imaging ◽  
Cost Issue

Reflectivity images obtained by prestack depth migration are often distorted by uneven subsurface illumination, especially in areas with complex geology, such as subsalt regions. We address the problem of uneven illumination in subsalt imaging by posing the reflectivity-imaging problem as a linear inverse problem and solving it in the image domain in a target-oriented fashion. The most computationally intensive part of the image-domain inversion is the explicit computation of the so-called Hessian operator. The Hessian is defined to be the normal operator of the associated modeling/imaging operator, which is a direct measure of the illumination deficiency of the imaging system. We can overcome the cost issue by using the phase-encoding technique in the 3D conical-wave domain for marine streamer acquisitions. We apply the inversion-based imaging methodology to a 3D field data set acquired from the Gulf of Mexico, and we precondition the inversion with nonstationary dip filters, which naturally incorporate interpreted geologic information. Numerical examples demonstrate that imaging by regularized inversion successfully recovers the reflectivities from the effects of uneven illumination, yielding images with more balanced amplitudes and higher spatial resolution.

A Boltzmann machine for high-resolution prestack seismic inversion

2019 ◽  
Vol 7 (3) ◽  
pp. SE215-SE224 ◽  
Author(s):  
Son Dang Thai Phan ◽  
Mrinal K. Sen
Keyword(s):  
Neural Network ◽  
High Resolution ◽  
Learning Algorithm ◽  
Hopfield Neural Network ◽  
Seismic Inversion ◽  
Good Prospect ◽  
Boltzmann Machine ◽  
Inversion Algorithm ◽  
Data Set ◽  
Prestack Seismic Inversion

Seismic inversion is one popular approach that aims at predicting some indicative properties to support the geologic interpretation process. Existing inversion techniques indicate weaknesses when dealing with complex geologic area, where the uncertainties arise from the guiding model, which are provided by the interpreters. We have developed a prestack seismic inversion algorithm using a machine-learning algorithm called the Boltzmann machine. Unlike common inversion approaches, this stochastic neural network does not require a starting model at the beginning of the process to guide the solution; however, low-frequency models are required to convert the inversion-derived reflectivity terms to the absolute elastic P- and S-impedance as well as density. Our algorithm incorporates a single-layer Hopfield neural network whose neurons can be treated as the desired reflectivity terms. The optimization process seeks the global minimum solution by combining the network with a stochastic model update from the mean-field annealing algorithm. Also, we use a Z-shaped sample sorting scheme and the first-order Tikhonov regularization to improve the lateral continuity of the results and to stabilize the inversion process. The algorithm is applied to a field 2D data set to invert for high-resolution indicative P- and S-impedance sections to better capture some features away from the reservoir zone. The resulting models are strongly supported by the well results and reveal some realistic features that are not clearly displayed in the model-based deterministic inversion result. In combination with well-log analyses, the new features appear to be a good prospect for hydrocarbon.

Sign in / Sign up

Export Citation Format

Share Document