Conference PaperPDF Available

CSEM exploration in the Barents Sea: Joint CSEM & seismic interpretation

Authors:
CSEM exploration in the Barents Sea: Joint CSEM & Seismic interpretation
Janniche Iren Nordskag*, Øyvind Kjøsnes, Anh Kiet Nguyen, and Ketil Hokstad, Statoil ASA
Summary
While seismic has a superior structural resolution and is
sensitive to porosity and the presence of hydrocarbons, it
has very little sensitivity to saturation. Controlled-source
electromagnetic (CSEM) is, on the other hand, sensitive to
hydrocarbon saturation. Thus, joint interpretation of
seismic and CSEM inversion results offer a great potential
for hydrocarbon saturation and volume estimations. We
describe a new workflow employing high resolution
deterministic CSEM inversion and statistical seismic AVO
inversion that applies in an exploration setting, i.e. without
prior assumptions of the background resistivities or the
existence, sizes and shapes of resistive anomalies. The
high resolution deterministic CSEM inversion has proven
better depth placements and more confined resistive
anomalies by incorporating seismic horizons into the
inversion. We use a statistical AVO inversion for lithology
and fluid predictions, called PCube, where a prior model
for the elastic properties is defined for different lithology
and fluid classes (LFCs) based on geological knowledge
and/or information from nearby wells. PCube outputs
posterior probability for each LFC specified in the prior
model. The inversion results from CSEM and seismic are
then co-interpreted utilizing their complementary
information. This offers a direct comparison between
expected hydrocarbon-filled reservoirs from seismic and
whether these reservoirs are of high or low saturation from
CSEM inversion results, and to explain resistive anomalies
not related to hydrocarbons. Applying the workflow on
Skrugard datasets, the joint interpretation clearly shows the
ability to discriminate high hydrocarbon saturated from
brine/low saturated sand reservoirs. Furthermore, the
workflow can also distinguish hydrocarbon related from
non-hydrocarbon related resistive anomalies.
Introduction
In hydrocarbon exploration, seismic has been the data of
choice for interpreters because of high spatial resolution,
its sensitivity to porosity and to the presence of
hydrocarbons, for instance in the form of direct
hydrocarbon indicators (DHIs) like flatspots. Although
seismic might be able to detect the presence of
hydrocarbons, it is not very sensitive to the fluid saturation.
Another hydrocarbon exploration tool that has been used
commercially in the last decade is controlled-source
electromagnetic, CSEM (Eidesmo et al., 2002), which is
sensitive to the saturation of saline water in the formation
pore space. Marine CSEM has since its conception evolved
immensely both on the hardware (more sensitive receivers
and powerful source), acquisition (3D grid) and processing
(3D anisotropic inversion) side towards better accuracy
and reduced interpretation ambiguity. Today, 3D CSEM
anisotropic inversions are commonly used to give a
resistivity model of the subsurface
Thus, joint interpretation of seismic and CSEM inversion
results offer a great potential for hydrocarbon saturation
and volume estimations (Harris et al., 2009 and Morten et
al., 2012). However, there are issues related to the poor
spatial and burial depth resolution of 3D CSEM inversion
results, see for example typical state-of-the-art results in
Gabrielsen et al. (2013). This is related to the low
frequency content in the source and to the fact that CSEM
is sensitive to the thickness-resistivity product, transverse
resistance (Constable and Weiss, 2006)
In seismic AVO analysis the interpreter is searching for
high porous sands and direct hydrocarbon indications, but
it might be hard to differentiate between a high and low
saturation case. Seismic interpretation may then be aided
by CSEM to look for prospects that have high saturation
and thereby large volumes.
The low resolution in CSEM inversion results complicates
joint interpretation/inversion with other geophysical
methods of much higher resolution such as seismic and
well logging. By including constraints in the CSEM
inversions, we might reduce the ambiguity and provide a
better resistivity image of the subsurface (Harris et al.,
2009 and Morten et al., 2012). However, constrained
CSEM inversions often require precise knowledge of the
background resistivities together with the size and shape of
all the anomalies to be studied. This prior information is
usually not available in an exploration setting.
Although CSEM data is sensitive to the water saturation in
subsurface formations and will give a resistive anomaly in
the case of high hydrocarbon saturation, other lithologies
might also give rise to resistivity anomalies, e.g. salt,
basalt, mature source rocks and consolidated sand. In
addition to this, CSEM data does not have the resolution to
differentiate between stacks of gas and oil saturated sand.
Joint interpretation with seismic inversion methods may
help resolving these ambiguities.
In this abstract we will present a workflow and a real data
example where high resolution CSEM inversion (Nguyen
et al., 2013) is interpreted jointly with a statistical seismic
AVO inversion (Buland and Omre, 2003), called PCube, to
DOI http://dx.doi.org/10.1190/segam2013-0883.1© 2013 SEG
SEG Houston 2013 Annual Meeting Page 810
Downloaded 04/06/16 to 143.97.2.35. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Joint CSEM & Seismic interpretation
reduce the uncertainties in the hydrocarbon volume
estimation and to discriminate between high hydrocarbon
saturated sand from other causes of resistive anomalies.
This study is a part of a larger study that includes CSEM
inversions, triaxial induction log measurements and
seismic inversions of different Skrugard datasets (Løseth et
al., 2013 and Nguyen et al., 2013).
Seismic AVO inversion, PCube
The seismic inversion is a linearized Bayesian AVO
inversion for lithology and fluid predictions called
Probability Cube (PCube) (Buland and Omre, 2003 and
Buland et al., 2008).
To compute a prior model for the elastic properties, one
defines lithology and fluid classes (LFCs) based on
geological knowledge and/or nearby well log(s). The rock
physics properties of each LFC are then specified by a
likelihood function and each LFC are given a probability
distribution, see Figure 1. Typical LFCs are different types
of shale and sand, where the sands are also defined with
different fluids (brine, oil and gas). Nearby well log(s) are
also used to estimate a proper source wavelet which quality
is essential for the inversion results.
Outputs from PCube are spatial probability cubes for each
of the LCFs defined in the prior model and an undefined
cube where the seismic data does not fit any of the
classified LFCs.
Since CSEM can typically not discriminate between oil
and gas, the posterior probability cube for oil and gas
saturated sands are merged into one cube that gives the
posterior probability for hydrocarbon saturated sandstone.
High resolution CSEM inversion
The high resolution deterministic CSEM inversion is
described in (Nguyen et al., 2013). In short, it is a
workflow including a stitched common midpoint (CMP)
anisotropic inversion utilizing seismic horizons to create
initial models for a pixel based, anisotropic 3D inversion
tool. Since no assumptions about the existence, size and
shape of reservoirs are made in the workflow, it is
applicable in a frontier exploration setting. The workflow
for the 3D inversion described by Nguyen et al. (2013)
gives better placements and confinements in depth for
resistive anomalies.
Different from the CSEM inversion presented in Nguyen et
al. (2013) for the 2011 CSEM dataset, we have here two
3D CSEM inversions, one for the southern part and one for
the northern part. The reason for this split up is the
different depth of targets in the subsurface. We use
frequencies f=1,2,4Hz and a much finer inversion and
modeling grids dx=dy=100m and dz=40m for the northern
part, due to shallow targets. Higher frequencies and denser
but smaller numerical grids ensure higher resolutions in the
inverted resistivity models. In the 3D CSEM inversion for
the northern part, we have used a smoothing regularization
down to a certain formation in order to mute receiver
imprints. For the southern part we use frequencies f = 1, 2
Hz, and the modeling and inversion grid is dx=dy=150 m
and dz=40 m. The final inverted models have data misfit
below 5%.
Joint interpretation
When the inversion results from seismic and CSEM are
satisfactory, they are co-visualized and co-interpreted
using appropriate software. By utilizing the
complementary information from PCube and 3D CSEM
inversion, we are able to discriminate between high and
low hydrocarbon saturation for a reservoir, and possibly
explain resistive anomalies not related to hydrocarbon
reservoirs.
Example
We apply the workflow on the 2011 CSEM Skrugard
dataset and a seismic multi-client dataset acquired by
WesternGeco in 2008. After running the CSEM and
seismic inversions separately, they are co-visualized for
joint interpretation.
Figure 2 shows sections of the inverted vertical resistivity
over the Skrugard discovery and the 7219/9-1 well with
residual hydrocarbon saturation, which was drilled by
Norsk Hydro in 1988. It is a clear resistive anomaly
precisely bounded by Skrugard top reservoir and the oil-
water contact. Note that no assumptions about Skrugard or
the background resistivities have been made in the
inversion. Precisely at the location of the 7219/9-1 well, on
the other hand, no resistive anomaly is found. To the right
for the 7219/9-1well, there is a resistive anomaly which
origin is at present not determined.
Figure 3 shows the PCube a posteriori probability for
hydrocarbon saturated sand over the same regions. PCube
predicts high probability for hydrocarbon saturated sand at
Skrugard. The predicted volume fits well with top reservoir
and second flat spot. However, PCube also predicts high
probability for hydrocarbons in a formation underneath the
sand formation where the Skrugard discovery was made
and the 7219/9-1 well was tested and found only residual
hydrocarbon saturation. Here, the inverted CSEM models
do not have any anomalies indicating that the hydrocarbon
saturation is low or zero.
DOI http://dx.doi.org/10.1190/segam2013-0883.1© 2013 SEG
SEG Houston 2013 Annual Meeting Page 811
Downloaded 04/06/16 to 143.97.2.35. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Joint CSEM & Seismic interpretation
On the other hand, PCube provides low/no probability for
hydrocarbon saturated sand at the location of the resistive
anomaly to the right for the 7219/9-1 well that indicates
the origin of this anomaly is perhaps not hydrocarbon
related.
Conclusion
We describe a new workflow combining deterministic high
resolution CSEM inversion and statistical seismic AVO
inversion and apply it on Skrugard CSEM and seismic
datasets. The work flow is applicable in an exploration
setting. The joint interpretation clearly shows the ability to
discriminate high hydrocarbon saturated from brine/low
saturated reservoirs. Furthermore, the workflow can also
distinguish hydrocarbon related from non-hydrocarbon
related resistive anomalies.
Acknowledgements
We thank the Skrugard Exploration Team for valuable
inputs and cooperation. We also thank Statoil and the
Skrugard license partners (Eni Norge and Petoro), and
WesternGeco for permission to publish the results.
Figure 1: Acoustic impedance versus P- and S-wave velocity ratio for the classified facies in PCube using the Skrugard well log.
DOI http://dx.doi.org/10.1190/segam2013-0883.1© 2013 SEG
SEG Houston 2013 Annual Meeting Page 812
Downloaded 04/06/16 to 143.97.2.35. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Joint CSEM & Seismic interpretation
Figure 2: Vertical resistivity section through the Skrugard well (left panel) and the 7219/9-1 well (right panel) of the 3D CSEM
inversion result overlain on depth migrated seismic.
Figure 3: Posterior hydrocarbon probability from PCube through the Skrugard well (left panel) and the 7219/9-1 well (right
panel). The bright yellow colour shows high probability for hydrocarbon saturated sand. Water surface multiples contaminate
the seismic above the solid black line invalidating the PCube results here.
DOI http://dx.doi.org/10.1190/segam2013-0883.1© 2013 SEG
SEG Houston 2013 Annual Meeting Page 813
Downloaded 04/06/16 to 143.97.2.35. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
http://dx.doi.org/10.1190/segam2013-0883.1
EDITED REFERENCES
Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2013
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for
each paper will achieve a high degree of linking to cited sources that appear on the Web.
REFERENCES
Buland, A., and H. Omre, 2003, Bayesian linearized AVO inversion: Geophysics, 68, 185198,
http://dx.doi.org/10.1190/1.1543206.
Buland, H., O. Kolbjørnsen, R. Hauge, Ø. Skjæveland, and K. Duffaut, 2008, Bayesian lithology and
fluid prediction from seismic prestack data: Geophysics, 73, no. 3, C13C21,
http://dx.doi.org/10.1190/1.2842150.
Constable, S. , and C. J. Weiss, 2006, Mapping thin resistors and hydrocarbons with marine EM methods:
Insights from 1D modeling: Geophysics, 71, no. 2, G43G51, http://dx.doi.org/10.1190/1.2187748.
Eidesmo, T., S. Ellingsrud, L. M. MacGregor, S. Constable, M. C. Sinha, S. Johansen, F. N. Kong, and H.
Westerdahl, 2002, Sea Bed Logging (SBL), a new method for remote and direct identification of
hydrocarbon filled layers in deepwater areas: First Break, 20, 144152.
Gabrielsen, P. T., P. Abrahamson, M. Panzner, S. Fanavoll, and S. Ellingsrud, 2013, Exploring frontier
areas using 2D seismic and 3D CSEM data, as exemplified by multi-client data over the Skrugard
and Havis discoveries in the Barents Sea: First Break, 31, 6673.
Harris, P. E., Z. Du, L. MacGregor, W. Olsen, R. Shu, and R. Cooper, 2009, Joint interpretation of
seismic and CSEM data using well log constraints: an example from the Luva Field: First Break, 27,
7381.
Løseth, L. O., T. Wiik, P. A. Olsen, A. Becht, and J. O. Hansen, 2013: CSEM in the Barents Sea, Part I:
Detecting Skrugard from CSEM: 75th Annual International Meeting, EAGE, Expanded Abstracts,
68566.
Morten, J. P., F. Roth, S. Karlsen, D. Timko, C. Pacurar, P. A. Olsen, A. K. Nguyen, and J. Gjengedal,
2012, Field appraisal and accurate resource estimation from 3D quantitative interpretation of seismic
and CSEM data: The Leading Edge, 31, 447456, http://dx.doi.org/10.1190/tle31040447.1.
Nguyen, A. K., Ø. Kjøsnes, and J. O. Hansen, 2013: CSEM in the Barents Sea, Part II: High resolution
CSEM inversion: 75th Annual International Meeting, EAGE, Expanded Abstracts, 68567.
DOI http://dx.doi.org/10.1190/segam2013-0883.1© 2013 SEG
SEG Houston 2013 Annual Meeting Page 814
Downloaded 04/06/16 to 143.97.2.35. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
Article
Non‐linear inversion of controlled source electromagnetic data is non‐unique. Inversion ambiguity and uncertainty grow with model complexity and limitations in sensitivity, for example when imaging deep targets. Uncertainties should be estimated. The best way to do that is by statistical inversion techniques. Because of the large number of parameters of 3D models, these methods are too costly for 3D applications. But neglecting or underestimating uncertainties often leads to erroneous interpretation and poor exploration decisions. The number of inversion parameters can be dramatically reduced while still being able to describe a complex 3D subsurface. To obtain that, I represent prior information by an input ensemble of geologically reasonable prior models. The model space is defined by the principal directions of this ensemble, where the inversion parameters are the coefficients of the different components. In this model space, inversion becomes much more efficient and can produce an updated ensemble of posterior models at the same computational cost as a controlled‐source electromagnetic inversion project using conventional non‐linear inversion techniques. I present examples where unique Gauss–Newton inversions lead to difficult interpretations and poor conclusions about hydrocarbon presence and saturation. The new approach provides a quantitative estimation of resistivity ambiguities and uncertainties leading to better interpretation and safer decisions. The examples use 2D synthetic controlled‐source electromagnetic data and 3D controlled‐source electromagnetic field data.
Chapter
The most utilized technique for exploring the Earth's subsurface for petroleum is reflection seismology. However, a sole focus on reflection seismology often misses opportunities to integrate other geophysical techniques such as gravity, magnetic, resistivity, and other seismicity techniques, which have tended to be used in isolation and by specialist teams. There is now growing appreciation that these technologies used in combination with reflection seismology can produce more accurate images of the subsurface. This book describes how these different field techniques can be used individually and in combination with each other and with seismic reflection data. World leading experts present chapters covering different techniques and describe when, where, and how to apply them to improve petroleum exploration and production. It also explores the use of such techniques in monitoring CO2 storage reservoirs. Including case studies throughout, it will be an invaluable resource for petroleum industry professionals, advanced students, and researchers.
Article
Full-text available
In this paper we describe a technique called Sea Bed Logging (SBL), an application of marine CSEM sounding, which can be applied to detect and characterize hydrocarbon bearing reservoirs in deep water areas. The basis of the approach is the use of a mobile horizontal electric dipole (HED) source and an array of seafloor electric field receivers. The transmitting dipole emits a low frequency electromagnetic signal both into the overlying water column and downwards into the seabed. The array of sea floor receivers measures both the amplitude and the phase of the received signal that depend on the resistivity structure beneath the seabed. A survey consisting of many transmitter and receiver locations can be used to determine a multidimensional model of subseafloor resistivity. In deep water areas the geological strata are generally dominated by shale or mud rocks with rather low resistivity. A hydrocarbon reservoir can have resistivity perhaps 10-100 times greater. With an in-line antenna configuration the transmitted electric field enters the high resistive hydrocarbon layer under a critical angle and is guided along the layer. Electromagnetic signals constantly leak from the layer and back to the seafloor. The guiding of the electric fields significantly alters the overall pattern of current flow in the overburden layer. A broad-line antenna configuration does not generate guided waves, and thus the two antenna configurations have different sensitivity to thin buried resistive layers. This so-called 'split' effect that is diagnostic for buried resistive layers is verified by 1-dimensional modelling, 1- and 2-dimensional modelling and real data acquired offshore West Africa also demonstrate that by careful positioning of transmitter tow tracks and receivers relative to a suspected hydrocarbon bearing structure, the SBL technique can provide detailed information on the presence and lateral extent of the hydrocarbon reservoir.
Article
Full-text available
The use of marine controlled-source electromagnetic EM (CSEM) sounding to detect thin resistive layers at depths below the seafloor has been exploited re-cently to assess the resistivity of potential hydrocar-bon reservoirs before drilling. We examine the sensitiv-ity of the CSEM method to such layers with forward and inverse modeling in one and three dimensions. The 3D modeling demonstrates that if both source and receivers are over a tabular 3D target, 1D modeling predicts the observed response to very high accuracy. Experimental design can thus be based on 1D analy-sis in which hundreds of range and frequency combi-nations can be computed to find the optimal survey parameters for a given target structure. Modeling in three dimensions shows that the vertical electric-field response is largest over the edges of a 3D target. The 3D modeling also suggests that a target body needs to have a diameter twice the burial depth to be reli-ably seen by CSEM sounding. A simple air-wave model (energy propagating from source to receiver via the atmosphere) allows the effects of the target layer and atmosphere to be separated and shows where sensi-tivity to the target is diminished or lost because of fi-nite water depth as a function of range, frequency, and seafloor resistivity. Unlike DC resistivity sounding, the marine CSEM method is not completely T-equivalent and, in principle, can resolve resistivity and thick-ness separately. Smooth inversion provides an estimate of the method's resolving power and highlights the fact that although the radial CSEM fields contain most of the sensitivity to the thin resistive target, inverted alone they produce only increasing resistivity with depth. In-clusion of the radial mode CSEM data forces the recov-ery of the thin resistor, but magnetotelluric data can be used more effectively to achieve the same result. Manuscript received by the Editor December 1, 2004; revised manuscript received June 28, 2005; published online March 9, 2006.
Article
There is new and growing enthusiasm for hydrocarbon exploration in the Barents Sea after three recent discoveries: Skrugard, Havis, and Norvarg. We demonstrate how using wide-azimuth 3D controlled-source electromagnetic (CSEM) and 2D seismic data together can improve the identification of prospective areas in the region. An interpretation workflow is presented to show how the challenging resistivity background in the Barents Sea is handled. To do this, we introduce a new inversion attribute called the anomalous vertical resistivity. The inversion attribute is co-visualized with 2D seismic data and used to estimate recoverable reserves. The workflow is illustrated on both synthetic and real data for the Skrugard and Havis discoveries. Both discoveries are identified on CSEM maps and by integrating CSEM data with seismic data. In addition, a new lead is identified. We show that CSEM data carries structural information and that the horizontal resistivity trends can be used with 2D seismic data and well logs to interpret the distribution of good-quality sands. Hydrocarbon reserve calculations based on the 3D CSEM data for the Skrugard and Havis discoveries have P50 values consistent with the publicly available reserve estimates. The reserve estimate for the new lead also shows significant potential.
Article
The key questions in field appraisal are: What is the hydrocarbon volume, and how are the hydrocarbons distributed in the field? The ability to answer these questions accurately is critical for deciding whether to produce a field and for developing a production plan. Wells drilled during the appraisal phase provide well and flow-test data, which are combined with structural knowledge from seismic surveys to map the extent of the field and generate a reservoir model. The cost for appraising an offshore field can exceed US $100 million, and it is desirable to obtain the information required with fewer wells if possible. Quantitative interpretation of surface geophysical data provides reservoir properties between well locations and can, therefore, significantly reduce appraisal costs. A quantitative analysis of seismic data using well-log information will typically determine reservoir rock porosity. Other important parameters are hydrocarbon saturation, permeability, and net-to-gross ratio. Quantitative interpretation of several reservoir properties using only the seismic data is associated with significant ambiguity. To determine several of these parameters accurately, complementary geophysical data sets with different sensitivity characteristics are needed. Controlled-source electromagnetic (CSEM) data are sensitive to the fluid type and can provide additional information to determine the hydrocarbon saturation more accurately. We have developed a new quantitative interpretation workflow integrating seismic and marine 3D CSEM data for estimating the hydrocarbon volume and obtaining 3D distributions of the hydrocarbon pore volume. A performance test of the workflow has been carried out on the Troll West Oil Province (TWOP) in the Norwegian North Sea (Figures 1 and 2). This article describes our methodology and presents encouraging results.
Article
A fast Bayesian inversion method for 3D lithology and fluid prediction from prestack seismic data, and a corresponding feasibility analysis were developed and tested on a real data set. The objective of the inversion is to find the probabilities for different lithology-fluid classes from seismic data and geologic knowledge. The method combines stochastic rock physics relations between the elastic parameters and the different lithology-fluid classes with the results from a fast Bayesian seismic simultaneous inversion from seismic data to elastic parameters. A method for feasibility analysis predicts the expected modification of the prior probabilities to posterior probabilities for the different lithology-fluid classes. The feasibility analysis can be carried out before the seismic data are analyzed. Both the feasibility method and the seismic lithology-fluid probability inversion were applied to it prospect offshore Norway. The analysis improves the probability for gas sand from 0.1 to about 0.2-0.4 with seismic data.
Article
A new linearized AVO inversion technique is developed in a Bayesian framework. The objective is to obtain posterior distributions for P-wave velocity, S-wave velocity, and density. Distributions for other elastic parameters can also be assessed-for example, acoustic impedance, shear impedance, and P-wave to S-wave velocity ratio. The inversion algorithm is based on the convolutional model and a linearized weak contrast approximation of the Zoeppritz equation. The solution is represented by a Gaussian posterior distribution with explicit expressions for the posterior expectation and covariance; hence, exact prediction intervals for the inverted parameters can be computed under the specified model. The explicit analytical form of the posterior distribution provides a computationally fast inversion method. Tests on synthetic data show that all inverted parameters were almost perfectly retrieved when the noise approached zero, With realistic noise levels, acoustic impedance was the best determined parameter, while the inversion provided practically no information about the density. The inversion algorithm has also been tested on a real 3-D data set from the Sleipner field. The results show good agreement with well logs, but the uncertainty is high.
2013: CSEM in the Barents Sea, Part II: High resolution CSEM inversion
  • A K Nguyen
  • Ø Kjøsnes
  • J O Hansen
Nguyen, A. K., Ø. Kjøsnes, and J. O. Hansen, 2013: CSEM in the Barents Sea, Part II: High resolution CSEM inversion: 75th Annual International Meeting, EAGE, Expanded Abstracts, 68567.
This reference list is a copy-edited version of the reference list submitted by the author
  • References Note
REFERENCES Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2013
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web
  • A References Buland
  • H Omre
SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web. REFERENCES Buland, A., and H. Omre, 2003, Bayesian linearized AVO inversion: Geophysics, 68, 185-198, http://dx.doi.org/10.1190/1.1543206.