Identifying the location of hydrocarbon reservoirs from

seismic data using continuous wavelet transform

Adel Hooshyar, Mohammad Ali Riahi, GHasem Zargar

Abadan Petroleum University of Technology (PUT),Abadan,Iran

MS student, Adel Hooshyar, Abadan Petroleum University of Technology
(PUT),Abadan,Iran,adel.hooshyar.nd@yahoo.com

Graduated in general Oil Engineering, Amirkabir university of Tehran in SEP., will be

graduated until in MSc Oil Exploration
Identifying the location of hydrocarbon reservoirs from
seismic data using continuous wavelet transform

Abstract:
It is decades from beginning of usage of seismic method and the last three decades is the
peak of efforts and studies to improve the efficiency of the method for displaying the better
and more accurate picture from complex geological structures. In seismic sections many
events and important geological structures in reservoir discussions are not recognizable and
are invisible in usual methods, so lots of efforts for using different approaches and
innovation of new methods are done to determine some other important properties that are
concerned in geology and petroleum engineering.
Access to reservoir geology, reservoir fluid and the reservoir geometry from seismic data,
using different analytical methods, is one of the most challenges in petroleum engineering.
However, conventional seismic techniques before were not successful in determining
and drawing on thin sections and indicating hydrocarbon locations. Spectral decomposition
methods were great progress in determining reservoirs characteristics. Also, complex
reservoir structures such as reef that have a significant potential of hydrocarbon
accumulation, determine the low-frequency shadows beneath hydrocarbon reservoirs,
especially in important gas reservoirs including sandstone reservoir that is not visible in
usual and common seismic sections and several other items.Since the time-frequency
mapping is a non-unique process, there are many approaches for non-static signal analysis.
In the past two decades Wavelet Transform has been used in many branches of science and
engineering. Continuous wavelet transform (CWT) uses a different method in time-
frequency analysis. In this way, instead of supplying time-frequency spectrum, it provides
time-scale map which is called Scalogram. Since the scale shows a frequency band, it is
understandable to interpret signals frequency content.
Spectral analysis methods can be powerful approaches of extracting information from
seismic data to reach the location of hydrocarbon reservoir. In recent efforts, Spectral
analysis methods have been introduced as a direct hydrocarbon indicator. This is paper
shows the application of CWT as a hydrocarbon indicator to reveal the location of gas in a
southern Iranian field.

Keywords: direct hydrocarbon indicator, seismic, spectral decomposition, time-

frequency mapping, Continuous wavelet transform

. Introduction:

In seismic exploration, all the methods that provide continuous analysis of time -
frequency of seismic trace is called spectral analysis. So it is possible to have the
frequency spectrum for each temporal sample of seismic traces. Figure -a is an
example of trace with different frequencies and figure -b is its frequency spectrum [].
In the early nineties the method was promoted to one of the most important tool for
seismic data interpretation. So far, for the main applications, this technique has
concentrated to successfully detect the subtle stratigraphic features [,] and recently as
a direct hydrocarbon indicator is used.
Figure-a) Trace that is made by different frequencies superimposed un each other and -b) its frequency
spectrum.

From the beginning of digital recording, the seismic trace was decomposed to its
Fourier components to suppress the low frequency ground rolls, the noises from human
activities, and high frequency random noises. Many features that are difficult to
visualize in the time domain often can be illustrates in frequency-domain representation
of a time-series. The manner in which the time series is mapped into the frequency
domain determines the amount of new information that can be obtained. The amplitude
and phase spectra of a whole seismogram represent the frequency behavior averaged
over the entire (stationary) time series. This standard Fourier-based method is a tool for
interpretation and is the basis for seismic data processing methods such as frequency
filtering, deconvolution, and wavelet characteristics. Alternatively, instantaneous
frequency and phase are attributes of complex trace analysis that are used to describe
changes in spectral behavior along the seismogram. However, these attributes are scalar
parameters and do not describe the complete spectrum at each time point [].
Since seventies the spectral balancing technique and spectral filtering with moving
window was began. In this way, each trace divides to series of shorter overlap traces;
those are concentrating temporally around events in original trace. The longer window,
usually between milliseconds and milliseconds, makes better quality in
frequency resolution; but shorter window with length milliseconds, provides
weaker results. This shortage is result of an assumption that is a reflection of lower
layers has a white spectrum. But the reflections from lower layers are not white and this
will distribute the spectral balancing during the process. The seismic data are not
stationary in nature so their frequency content will change during the time. According to
spectral balancing, the spectrum analysis of shorter windows was discussed since
middle of Nineties [].
Usually the seismograms with highly changeable frequency content considered a non-
stationary signal so they need Non-standard methods for decomposition. For a more
complete description of the frequency content varies with the time, it needs the spectral
decomposition of signal in two dimensional time-frequency domain. In this way the
entire bandwidth spectrum for each temporal sample can be described and it can be used
to detect superimposed seismic events.Idea of using spectral analysis with shorter
windows comes from seismic interpretation, not from seismic processing. Seismic
Interpreters are interested in related spectral measurements, such as measuring the
frequency where the tuning occurs, and do not require absolute frequency content.
The content of this spectrum can be used to detect lateral changes in thickness and
homogeneous, also it can be used as a direct hydrocarbon Indicator.
Most of functions and their transforms are considered into two domains. In the old text,
they refer to high and low domains. They also named as a function domains and
transform domains but in most physical applications these areas are called time and
frequency domains.
Since time-frequency mapping is a non-unique process, there exist various methods for
time-frequency analysis of non-stationary signals.So from a single seismic trace can
produce various time-frequency analyses. There are a variety of spectral decomposition
methods. These include the DFT (discrete Fourier Transform), MEM (maximum
entropy method), DWT (discrete wavelet transform), CWT (continuous wavelet
transform), and MPD (matching pursuit decomposition). None of these methods are,
strictly speaking, right or wrong. Each method has its own advantages and
disadvantages, and different applications require different methods [].
Wavelet analysis, determines frequency distribution of a non-stationary time series
using a series of windows which are partial in time support and band limited in
frequency domain. This window functions are similar to a small wave which find grow
or decrease in short periods of time, And therefore their names are wavelet. Methods
like wavelet shaping or wavelet deconvolution that are using usually in seismic data
processing, should not be confused with wavelet analysis that is a new published issue
in mathematics and signal processing. As same as Fourier analysis, the wavelet analysis
concludes transforms like wavelet series expansion, continuous wavelet transform,
discrete wavelet transform and wavelet packet transform. These transforms are all
reversible and therefore suitable for data filtering. Wavelet analysis acts like a
mathematical microscope. When a window function has a long length in time domain,
the roughness image of under studying signal structure will be obtained. When the scale
changes, i.e. the analyze window is narrow, Partial signal characteristics will be seen
well. The continuous wavelet transform first time was introduced by Morlet () and
Goupillaud (), but the union signal processing attention, was completely attracted
by Mallat ().

. Experimental:
 .. Continuous Wavelet Transform:
Continuous wavelet transform is an alternating method for analysis of frequency
distribution of non-stationary time series. The temporal signal decomposes to time-scale
spectrum that is called Scalogram. There are so many methods for transforming the
time-scale spectrum to time-frequency spectrum that are based on wavelet transform.
Wavelet analysis specifies the frequency distribution of non-stationary time series using
a set of windows.
The higher scales correspond to the most "stretched" wavelets. The more stretched the
wavelet, the longer the portion of the signal with which it is being compared, and thus
the coarser the signal features being measured by the wavelet coefficients. Thus, there is
a correspondence between wavelet scales and frequency as revealed by wavelet analysis
(figure ):
Low scale a=> Compressed wavelet =>Rapidly changing details => High frequency
High scale a=>Stretched wavelet=>Slowly changing, coarse features => Low
frequency

Figure : Comparison between low and high scale (Matlab, ).

The CWT is an alternative method to analyze a signal. In the CWT, wavelets dilate in
such a way that the time support changes for different frequencies. Smaller time support
increases the frequency support, which shifts toward higher frequencies. Similarly,
larger time support decreases the frequency support, which shifts toward lower
frequencies. Thus, when the time resolution increases, the frequency resolution
decreases, and vice versa [].
A wavelet is defined as a function ( ) L with a zero mean, localized in both time
and frequency. By dilating and translating this wavelet ( ), we produce a family of
wavelets:

( , )( )= ( ) ()

Where , and is not zero and is the dilation parameter or scale. Note that the
wavelet is normalized such that the L- norm is equal to unity. The CWT is
defined as the inner product of the family of wavelets ( , ) ( ) with the signal (t).
This is given by:

( , )= ( ), ( , )( ) = ( ) ()

Where is the complex conjugate of and ( , ) is the time-scale map (i.e., the
scalogram). The convolution integral in equation () can be computed easily in the
Fourier domain. The choice for the scale and the translation parameter can be arbitrary,
and we can choose to represent it any way we like. To reconstruct the function ( )
from the wavelet transform, we use Calderons identity, given by:

( )= ( , ) ( ) ()

For the inverse transform to exist, we require that the analyzing wavelet satisfy the
admissibility condition, given by:

( )
= < ()

Where ( ) is the Fourier transform of ( ) and where is a constant for wavelet .

The integrand in equation () has an integrable discontinuity at = and also implies
that ( ) = . A commonly used wavelet in continuous wavelet transform is the
Morlet wavelet, defined as :

( )= ( )

Where is the frequency and it is taken as to satisfy the admissibility condition.

The center frequency of the Morlet wavelet being inversely proportional to the scale
provides an easy interpretation from scale to frequency.

. . Frequency Attenuation:
The improved spectral analysis of seismic data can reveal hydrocarbon indications that
are not apparent on conventional seismic data. There are three distinct spectral
hydrocarbon indicators that are best revealed by proper spectral decomposition:
-seismic attenuation: The hydrocarbon reservoirs that filled with gas have interesting
behavior in the case of frequency dependent seismic attenuation. The usual approach to
measure this attenuation is to presume that the slope of ratio of frequency spectra for
two time window is directly related to the attenuation coefficient. We can use CWT to
display the instantaneous frequency to observe frequency dependent attenuation;
-low frequency shadows: These shadows are often attributed by explorationists to
abnormally high attenuation in gas-filled reservoirs. However, it is often difficult to
explain observed shadows under thin reservoirs where there is insufficient travel path
through absorbing gas reservoir to justify the observed shift of spectral energy from
high to low frequencies.
And -preferential illumination: The tuning thickness is the thickness at which a
reservoir is preferentially illuminated at a given dominant frequency. Since single
frequency seismic volumes can be obtained over any range of frequencies, there is not
one tuning thickness. Rather, there is a tuning frequency at which the target is
preferentially illuminated. This can help to have a proof. For a layer of constant
thickness, the tuning frequency will be different for brine and gas saturated rock and the
tuning frequency itself can be mapped as a hydrocarbon indicator.

.Results and Discussion:

The seismic cube of known gas reservoir of south field of Iran was taken (figure ).
This field has a known gas reservoir in - ms . First the CWT was operated on the
cube to have the instantaneous frequency (figure ). It was taken by industrial software.
Then the single frequency cubes was extracted from spectral cube. You can see the red
point that refers to HZ frequency in figure . These shadows are in - ms that the
gas reservoir in known in - ms by other exploration operations and wells. Figure
shows the frequency cube in HZ. It is clear that points in - ms in HZ
frequency are disappeared in HZ frequency.

Figure : The seismic cube of one of the southern Iranian oil field.

Figure : the instantaneous frequency cube of figure .

 Figure : the cube in HZ frequency (the low frequency shadows are clear in - ms).

Figure : the cube in Hz frequency (the low frequency shadows in HZ frequency is gone).

.Conclusions:

The wavelet transform decomposes the seismic trace to its constituent components. The
frequency spectrum of seismic trace is the summation of frequency spectrums of
wavelets that are constituent components of seismic trace, superposition principle said.
The frequency spectrum in each time is the result of superposition of weighted wavelet
spectrums that is in temporal neighborhood of sample. Continuous wavelet transform is
the most powerful operator between spectral decomposition methods in imaging the
instantaneous frequency of seismic cube.
The low frequency shadows are often results of gas reservoirs. It is well established that
gas-filled reservoirs exhibit higher frequency dependent seismic attenuation than similar
rock fully-saturated with brine.
It is clear from our results that CWT can help to determine the location of gas
reservoirs. The low shadows beneath the gas reservoir disappear in high frequencys.

.Acknowledgements:

The used data(seismic and logs) in this paper was provided by Iranian National
Exploration Management Company.

References:
[] Castagna, J.P., and Sun, S., , Comparison of spectral decomposition methods:
First Break, , .
[] Partyka, G., Gridley, J., and Lopez, J., , Interpretational applications of spectral
decomposition in reservoir characterization. The Leading Edge, , .
[] Sinha, S.K., Routh, R., Anno, P.D., and Castagna, J.P., , Spectral
decomposition of seismic data with continuous-wavelet transform: Geophysics, , -
.
[] Chakrabraty, A., and Okaya, D., , Frequency-time decomposition of seismic
data using wavelet-based methods: Geophysics, , -.
[] Liu, J., and Marfurt, K.J., , Instantaneous spectral attributes to detect channels.
Geophysics, (), -.
[]Mallat, S., , A wavelet tour of signal processing: Academic Press, nd edition.
.