Seogi Kang

SimPEG: An open source framework for simulation and gradient based parameter estimation in geophysical applications

Inverse modeling is a powerful tool for extracting information about the subsurface from geophysical data. Geophysical inverse problems are inherently multidisciplinary, requiring elements from the relevant physics, numerical simulation, and optimization, as well as knowledge of the geologic setting, and a comprehension of the interplay between all of these elements. The development and advancement of inversion methodologies can be enabled by a framework that supports experimentation, is flexible and extensible, and allows the knowledge generated to be captured and shared. The goal of this paper is to propose a framework that supports many different types of geophysical forward simulations and deterministic inverse problems. Additionally, we provide an open source implementation of this framework in Python called SimPEG (Simulation and Parameter Estimation in Geophysics, http://simpeg.xyz). Included in SimPEG are staggered grid, mimetic finite volume discretizations on a number of structured and semi-structured meshes, convex optimization programs, inversion routines, model parameterizations, useful utility codes, and interfaces to standard numerical solver packages. The framework and implementation are modular, allowing the user to explore, experiment with, and iterate over a variety of approaches to the inverse problem. SimPEG provides an extensible, documented, and well-tested framework for inverting many types of geophysical data and thereby helping to answer questions in geoscience applications. Throughout the paper we use a generic direct current resistivity problem to illustrate the framework and functionality of SimPEG.

Recovering IP information in airborne-time domain electromagnetic data

We propose a methodology to generate a 3D distribution of pseudo-chargeability from airborne time domain electromagnetic data. The processing flow is as follows: (a) Apply 3D inversion to TEM data to restore a background conductivity. This might involve omitting responses that are obviously contaminated with IP signals, such as negative transients in coincident loop surveys. The recovered background conductivity is assumed to be uncontaminated by IP signals. (b) Compute the TEM response from the background conductivity and subtract it from the observations. This yields the dIP data, and reduces the EM coupling. (c) The background conductivity is likely not exactly the earth conductivity, but we assume that the major effects of this inaccuracy will lead to a large scale, smoothly varying perturbation to the dIP data. If this correct, then these can be recognized and removed. (d) The final data are linearly related to a pseudo-chargeability through a sensitivity function that is analogous to that employed in usual DC-IP ground surveys. (e) The dIP data at various time channels can be inverted individually. The pseudo-chargeability models may be useful in themselves or they may be further processed to estimate Cole-Cole, or equivalent, parameters. We demonstrate our procedure on a field data set from Mt. Milligan. In the field example, we identify chargeable targets that show no indication of negative transients in the raw data. From the images we can make inferences about the relative strength and geometries of the chargeable bodies.

Inversion of airborne geophysics over the DO-27/DO-18 kimberlites — Part 2: Electromagnetics

AbstractWe focus on the task of finding a 3D conductivity structure for the DO-18 and DO-27 kimberlites, historically known as the Tli Kwi Cho (TKC) kimberlite complex in the Northwest Territories, Canada. Two airborne electromagnetic (EM) surveys are analyzed: a frequency-domain DIGHEM and a time-domain VTEM survey. Airborne time-domain data at TKC are particularly challenging because of the negative values that exist even at the earliest time channels. Heretofore, such data have not been inverted in three dimensions. In our analysis, we start by inverting frequency-domain data and positive VTEM data with a laterally constrained 1D inversion. This is important for assessing the noise levels associated with the data and for estimating the general conductivity structure. The analysis is then extended to a 3D inversion with our most recent optimized and parallelized inversion codes. We first address the issue about whether the conductivity anomaly is due to a shallow flat-lying conductor (associated with th...

mCSEM inversion for CO2 sequestration monitoring at a deep brine aquifer in a shallow sea

Carbon dioxide injection monitoring in offshore environments is a promising future application of the marine controlled-source electromagnetic (mCSEM) method. To investigate whether the mCSEM method can be used to quantitatively monitor variations in the distribution of the injected CO2, we developed a mCSEM inversion scheme and conducted numerical analyses. Furthermore, to demonstrate the monitoring capability of the mCSEM method in challenging environments, we used a deep brine aquifer model in shallow sea as an injection target. The mCSEM responses of the injected CO2 in the deep brine aquifer were severely decayed and heavily masked by the air wave due to the proximity of the free space. Therefore, the accurate computation of small mCSEM responses due to the injected CO2 and the proper incorporation into the inversion process are critically important for the mCSEM method to be successful. Additionally, in monitoring situations, some useful a priori information is usually available (e.g. well logs and seismic sections), and the proper implementation of this to our inversion framework is crucial to ensure reliable estimation of the distribution of the injected CO2 plume. In this study, we developed an efficient 2.5D mCSEM inversion algorithm based on an accurate forward modelling algorithm and the judicious incorporation of a priori information into our inversion scheme. The inversion scheme was tested with simplified and realistic CO2 injection models and successfully recovered the resistivity distributions of the injected CO2, although it still required the presence of a considerable amount of the injected CO2. Based on these inversion experiments, we demonstrated that the mCSEM method is capable of quantitatively monitoring variations in the distribution of injected CO2 in offshore environments. We developed an efficient 2.5D mCSEM inversion algorithm based on an accurate forward modelling algorithm and the judicious incorporation of a priori information into our inversion scheme. We demonstrated the successful recovery of resistivity distributions of the injected CO2 from the deep brine aquifer model in the shallow sea.

Inversion of airborne geophysics over the DO-27/DO-18 kimberlites — Part 3: Induced polarization

AbstractThe geologically distinct DO-27 and DO-18 kimberlites, often called the Tli Kwi Cho (TKC) kimberlites, have been used as a testbed for airborne geophysical methods applied to kimberlite exploration. This paper focuses on extracting chargeability information from time-domain electromagnetic (TEM) data. Three different TEM surveys, having similar coincident-loop geometry, have been carried out over TKC. Each records negative transients over the main kimberlite units and this is a signature of induced polarization (IP) effects. By applying a TEM-IP inversion workflow to a versatile time domain EM (VTEM) data set we decouple the EM and IP responses in the observations and then recover 3D pseudo-chargeability models at multiple times. A subsequent analysis is used to recover Cole-Cole parameters. Our models demonstrate that both DO-18 and DO-27 pipes are chargeable, but they have different Cole-Cole time constants: 110 and 1160 μs, respectively. At DO-27, we also distinguish between two adjacent kimber...

A framework for simulation and inversion in electromagnetics

Abstract Simulations and inversions of electromagnetic geophysical data are paramount for discerning meaningful information about the subsurface from these data. Depending on the nature of the source electromagnetic experiments may be classified as time-domain or frequency-domain. Multiple heterogeneous and sometimes anisotropic physical properties, including electrical conductivity and magnetic permeability, may need be considered in a simulation. Depending on what one wants to accomplish in an inversion, the parameters which one inverts for may be a voxel-based description of the earth or some parametric representation that must be mapped onto a simulation mesh. Each of these permutations of the electromagnetic problem has implications in a numerical implementation of the forward simulation as well as in the computation of the sensitivities, which are required when considering gradient-based inversions. This paper proposes a framework for organizing and implementing electromagnetic simulations and gradient-based inversions in a modular, extensible fashion. We take an object-oriented approach for defining and organizing each of the necessary elements in an electromagnetic simulation, including: the physical properties, sources, formulation of the discrete problem to be solved, the resulting fields and fluxes, and receivers used to sample to the electromagnetic responses. A corresponding implementation is provided as part of the open source simulation and parameter estimation project SimPEG ( http://simpeg.xyz ). The application of the framework is demonstrated through two synthetic examples and one field example. The first example shows the application of the common framework for 1D time domain and frequency domain inversions. The second is a field example that demonstrates a 1D inversion of electromagnetic data collected over the Bookpurnong Irrigation District in Australia. The final example is a 3D example which shows how the modular implementation is used to compute the sensitivity for a parametric model where a transmitter is positioned inside a steel cased well.

3D IP Inversion of Airborne EM data at Tli Kwi Cho

In this study, we revisit three airborne EM surveys over Tli Kwi Cho (TKC). These consist of a frequency domain DIGHEM data set, and two time domain surveys, VTEM and AeroTEM. Negative transients have been recorded in both of the time domain surveys and we interpret these as arising from chargeable bodies. The kimberlite pipes are referred to as DO-27 and DO-18. We look in more detail at the transient data and apply the ATEM-IP inversion procedure to recover a 3D pseudo-chargeability distribution. Important components of the analysis involve estimating a background conductivity for the region. For DO-27 we have used a 3D parametric inversion to recover the conductivity from TEM data. The IP signal for the inversion is obtained by subtracting the time domain responses estimated by EM inversion from the observed background signal. This process also removes EM coupling noise that might be contaminating the data. The resultant IP data are inverted with a linear inverse approach using the sensitivity from the background conductivity. This yields a 3D model of pseudo-chargeability.

Computation of Apparent Resistivity from Marine Controlled-source Electromagnetic Data for Identifying the Geometric Distribution of Gas Hydrate

요약: 해양 전자탐사의 겉보기 비저항은 해수층으로 인해 지표탐사와 그 정의가 달라지게 되며, 이를 적절히 계산할 수있는 알고리듬의 개발은 해양 전자탐사의 출발점이 될 수 있다. 이를 위해, 1차원 층서 가스 하이드레이트 수치모형과 해수층과 그 하부의 반 무한매질로 이루어진 수치모형에서 계산한 전자기적 반응을 비교분석하였다. 겉보기 비저항을 계산하기 위해서는 실수와 허수 성분 보다는 진폭과 위상을 사용하는 것이 더 적합하였으며 해양 전자탐사 반응의 민감도를정량적으로 분석하여, 근거리 영역에서는 위상이 원거리 영역에서는 진폭 성분이 더 안정적인 결과를 주는 것을 알았다.또한 위상과 진폭의 선택기준으로써 유도상수의 값을 제안하였다. 이러한 분석을 토대로 격자 탐색법(grid search)을 사용하여 겉보기 비저항을 계산하는 수치알고리듬을 개발하였다. 개발된 알고리듬을 이용하여 1차원 층서 가스 하이드레이트수치모형의 다양한 변수를 변화시켜가며 겉보기 비저항을 계산해봄으로써 알고리듬의 타당성을 검증하였다. 마지막으로,계산한 겉보기 비저항 값을 이용한 가스 하이드레이트 부존양상 정보의 도출가능성을 살펴보았다. 동해 울릉분지의 가스하이드레이트 부존양상을 모사한 2차원 가스 하이드레이트 수치모형에서 계산된 자료의 겉치레 단면도는 가스 하이드레이트 부존양상 정보 추출이 가능함을 보여주었다. 주요어: 해양 전자탐사, 겉보기 비저항, 가스 하이드레이트, 부존양상Abstract: The sea layer in marine Controlled-Source Electromagnetic (mCSEM) survey changes the conventionaldefinition of apparent resistivity which is used in the land CSEM survey. Thus, the development of a new algorithm,which computes apparent resistivity for mCSEM survey, can be an initiative of mCSEM data interpretation. First, wecompared and analyzed electromagnetic responses of the 1D stratified gas hydrate model and the half-space model belowthe sea layer. Amplitude and phase components showed proper results for computing apparent resistivity than real andimaginary components. Next, the amplitude component is more sensitive to the subsurface resistivity than the phasecomponent in far offset range and vice versa. We suggested the induction number as a selection criteria of amplitudeor phase component to calculate apparent resistivity. Based on our study, we have developed a numerical algorithm, whichcomputes appropriate apparent resistivity corresponding to measured mCSEM data using grid search method. In addition,we verified the validity of the developed algorithm by applying it to the stratified gas hydrate models with various modelparameters. Finally, by constructing apparent resistivity pseudo-section from the mCSEM responses with 2D numericalmodels simulating gas hydrate deposits in the Ulleung Basin, we confirmed that the apparent resistivity can provide theinformation on the geometric distribution of the gas hydrate deposit.Keywords: mCSEM, apparent resistivity, gas hydrate, geometrical distribution

TEM-IP: Extracting more induced polarization information from grounded source time domain electromagnetic data

Electrical induced polarization (EIP) surveys are used to detect chargeable materials in the earth. For interpretation of time domain EIP data a common procedure is to first invert the DC data (electric current on-time) to recover conductivity and then invert the IP data (current off-time) to recover chargeability. This DC-IP procedure assumes that the off-time data are free of secondary EM induction effects. To comply with this, early time data are often discarded, or not recorded. For mid-time data, an EM-decoupling technique, which removes EM induction in the observations, needs to be implemented. Usually responses from a half-space or a layered earth are subtracted. Recent capability in 3D time domain EM (TEM) forward modelling and inversion allows to revisit these procedures. In a TEM-IP survey, a high sampling rate allows early time channels of the EM data to be recorded. The recovery of chargeability then follows a 3-step workflow: (a) invert early time channel TEM data to recover the 3D conductivity; (b) use that conductivity to compute the TEM response at later time channels and subtract this fundamental response from the observations to extract the IP responses, and (c) invert the IP responses to recover a 3D chargeability. This workflow effectively removes EM induction effects in the observations and produces better chargeability and conductivity models compared to conventional approaches. In a synthetic example involving a gradient array we show that the conductivity structure obtained from the early time channel data, which are usually discarded, is superior to that obtained from the steady state DC voltages. This adds a further reason to collect these EM data. This article is protected by copyright. All rights reserved

Exploring nonlinear inversions: A 1D magnetotelluric example

At some point in many geophysical workflows, an inversion is a necessary step for answering the geoscientific question at hand, whether it is recovering a reflectivity series from a seismic trace in a deconvolution problem, finding a susceptibility model from magnetic data, or recovering conductivity from an electromagnetic survey. This is particularly true when working with data sets where it may not even be clear how to plot the data: 3D direct current resistivity and induced polarization surveys (it is not necessarily clear how to organize data into a pseudosection) or multicomponent data, such as electromagnetic data (we can measure three spatial components of electric and/or magnetic fields through time over a range of frequencies). Inversion is a tool for translating these data into a model we can interpret. The goal of the inversion is to find a “model” — some description of the earths physical properties — that is consistent with both the data and geologic knowledge.