POROSITY ESTIMATION FROM ELASTIC
PROPERTIES USING NEURAL NETWORK IN
CARBONATE RESERVOIR, CENTRAL LUCONIA
Sharanya Varatharajoo
Bachelor of Technology (Hons) Petroleum Geoscience
Universiti Teknologi Petronas
Bandar Seri Iskandar, Perak Darul Ridzuan
Email: sharanyarajoo@gmail.com
Abstract— Estimating porosity in carbonate reservoir could be
slightly difficult compared to siliciclastic due to the complex pore
system of carbonate rocks. In this study, the author estimate
porosity from elastic properties using neural network in carbonate
reservoir, Central Luconia Province, Sarawak. The aim of the
projects is (i) to predict the porosity in carbonates from seismic
inversion result using artificial neural network technology, (ii) to
evaluate the porosity estimation obtained from the empirical
equation and neural network with those from well log data to
check the accuracy, (iii) to assess the reliability of SQp and SQs
attribute for porosity prediction in carbonate body. Two type of
data such as well log data and seismic data were received. Elastic
properties were calculated, and cross-plots were created with
porosity versus calculated elastic properties. Post and pre-stack
inversion was conducted on near and far seismic data to extract
elastic properties. The determined elastic properties through
cross-plot and seismic inversion were used to create porosity
model through the empirical equation and neural network. Both
SQp and SQs is unfailing to predict porosity in carbonate
reservoir as it shows good trend in distinguishing the higher and
lower porosity values. However, SQs shows better correlation
(0.96662) and less inversion error (0.035), thus, it was selected as
an input for the empirical equation and neural network. Two
porosity models were created in this project. First model from
empirical equation of SQs and the second model from neural
network with inverted SQs as an input. Both the models were
compared with actual porosity from the well data to check its
accuracy and reliability. The porosity estimated from empirical
equation yielded slightly better correlation of 0.961911 compared
porosity estimated from neural network of 0.891984. Thus, SQs is
highly reliable to estimate the porosity in carbonate reservoir
through empirical equation and neural network.
Keywords- Elastic properties, Porosity, Neural Network, SQp and
SQs, Seismic inversion
INTRODUCTION
1.1 BACKGROUND OF STUDY
Predicting reservoir characterization such as porosity in
carbonates rocks can be difficult due to the complex pore
system of carbonates than siliciclastics. This is mainly due to
biological origin of carbonate sediment and its chemical
reactivity. Besides, the wide variety of grain shapes common in
carbonates, the presence of intragranular, framework, fenestral
carbonate porosity, and the potential for the development of
moldic and irregular dissolution- related carbonate porosity
also contributes to the complex pore system in carbonate rocks
[1].
The different depositional environment can be observed in
carbonates in Malaysia, which can be divided into four
subdivisions. This subdivision happens to be most suitable for
hands-on work, for example mapping of reservoir zones such
as protected, reefoid, shallow open marine offreef and deeper
open marine offreef.
In recent years, neural networks gained popularity among
geoscientist to predict the reservoir characterization [2].
Numerous research has been conducted and the results are quite
satisfying. Wong et al. [3] comments that based on a
nongenetic approach, the genetic approach developed good
porosity estimation in clastic rocks.
Neural network is also known as artificial neural network,
is a computing system technology vaguely inspired by
biological neural networks of human brains. Neural networks
basically eliminate or minimize the error in prediction and
processes [4]. Neural networks are being used in various field
where large data need to be processed. Handwriting recognition
for check processing, speech-to-text transcription, oil-
exploration data analysis, weather prediction and facial
recognition are several examples of using neural technology
network since 2000.
In this study it demonstrates the application of neutral
network to estimate porosity from elastic properties which
retrieved from the well log and seismic data. The aim of this
research is to implement neural network to solve porosity
estimation in carbonate reservoir of Field X in Central Luconia,
Malaysia.
1.2 PROBLEM STATEMENT
Porosity is the measurement of the void space in a rock that
enables to store fluids in it. Besides, it is also one of the
prominent factors in reservoir characterization together with
other properties such as permeability and water saturation.
Generally, the porosity can be predicted directly if we able to
retrieve an undamaged whole core from the reservoir. X-ray
computerized tomography scans, thin sections, and scanning
electron microscope examination will be used to determine the
porosity in the cores. Unfortunately, in many cases,
geoscientists failed to retrieve undamaged cores from the
reservoir. Thus, indirect predictions methods are widely being
used in the industry to estimate porosity outside of the well.
Few studies have been led to estimate the porosity by using
the electrical logs. The integration of well-log and seismic data
is a fundamental procedure for reservoir characterization. These
nonlinearity issues can be easily handled by using computer-
based intelligence methods such as artificial neural networks,
genetic algorithms and so on. Implementation of neural
network in determining porosity is commonly used in clastics
reservoirs and the reliability can already be justified from
numerous research. However, implementation of such
technology is rare in carbonates. In this project, porosity
estimation will be performed by applying neural network
technology. Nevertheless, can an accurate porosity estimation
will be produced in carbonates as the pore systems in
carbonates are much more complex than siliciclastic?
1.3 OBJECTIVES
There are several objectives that need to be achieved
at the end of the project. The objectives of the project are (1) to
estimate the porosity in carbonates body using artificial neural
network technology, (2) to evaluate the porosity estimation
obtained from the empirical equation and neural network with
those from well log data to check the accuracy and (3) to assess
the reliability of SQp and SQs attribute for porosity prediction
in carbonate body.
1.4 SCOPE OF STUDY
The research will be focusing in carbonate reservoir,
Central Luconia Province in Offshore Sarawak (Figure 1). In
this research, the well log and seismic data will be examined to
analyses the elastic properties. In addition, the elastic properties
will be calculated from well log data. While, elastic properties
also will be extracted from seismic data through pre-stack
inversion. Cross-plot analysis and inversion analysis will be
Figure 1: Location map of the Field X of Central Luconia,
Malaysia (Modified from [5])
determined which of the elastic properties give a good
correlation with porosity in carbonates. Later, the porosity
estimated from empirical equation and neural network using
HampsonRussell (Emerge) software. Lastly, the estimated
porosity will be compared with porosity from well data to check
the accuracy of the result.
LITERATURE REVIEW
2.1 TECTONIC HISTORY AND REGIONAL
STRUCTURAL SETTING OF THE SARAWAK BASIN
The tectonic evolution history of Borneo (Sabah, Sarawak,
and Kalimantan) is complicated and challenging for many
Geologists. Different Geologists have their own theory and
basically, this has been debated over more than two decades.
The regional Sunda Shelf which structurally connects Borneo
with Peninsular Malaysia and the rest of continental Southeast
Asia (Sundaland) includes the Sarawak continental margin.
Rifting of South China Sea, during the Eocene, the collision
of the accreted Luconia Shoals terrace with the West Borneo
Basement and the closure of Rajang Sea contributed to the
formation of the foreland basin of the southern part of the
Sarawak Basin. However, the collision takes places in an
anticlockwise rotational direction which resulted in forming a
series of horsetailed faults due to dextral strike-slip.
According to Tapponier [6], the collision of India Plate with
Eurasian Plate triggered the rifting of the South China Sea. This
led to the formation of micro-continental blocks. The Early
Paleocene led to the subduction of the Mesozoic oceanic crust
beneath the northern margin of the West Borneo Basement.
During the late Cretaceous to Eocene, the proto-South China
Sea (Rajang Sea) oceanic crust subducted southward beneath
the northern margin of the West Borneo Basement. The
Luconia Shoals Block collided and sutured with the West
Borneo Basement [7]. During Late Eocene to Early Oligocene,
due to the collision, a peripheral foreland basin was created at
the north of the Rajang Group collisional orogeny. The same
dextral forces rotated anti-clockwise and created
accommodation space for Cycle I and II sediments (end of
Cycle III times).
During Late Oligocene and Early Miocene, an anti-
clockwise extensional deformation occurred. Then a complex
regional unconformity, Middle Miocene Unconformity (MMU)
occurred with a time gap of 6-12 million years. Besides, due to
the compressional wrench deformation a severe inversion
occurred in the Cycle IV and younger deposition was
punctuated. The clastic transport to the North, diverting
sediments eastwards into the Baram Delta and westwards to the
Rajang Delta had been closed by Balingian Inversion [8].
During Late Miocene, the Luconia reefs grew uninterrupted.
2.2 NEURAL NETWORK TECHNOLOGY
Neural networks can solve for many variables at one time
also, can deal with the exceptionally complex statistical
analysis to make an estimation of lithology from huge seismic
information volumes [4]. Thus, neural networks are highly
recommended to geoscience related challenges.
In general, there are a wide range of neural networks. One
of the best known of neural networks is multilayer feed forward
(MLF), or back-propagation, neural network. The main
component of the MLF network is the neuron, which takes the
input, applies a nonlinear capacity to it, and creates an output.
This is known as a sigmoidal type (s-shaped) function. The
nonlinear nature can differentiate neural network from
multilinear regression, and can uncover the hidden information
in the data, however should be regarded with care so that the
false prediction can be overcome.
There are two main applications for neural networks in
seismic analysis. There are sample-to-sample prediction and
classification [4]. Firstly, let’s talk about sample-to sample
prediction where the objective is to distinguish each example
on a 3-D seismic volume with a lithological or liquid parameter,
for example, porosity or water saturation. The geoscientist
mainly utilized this method when they need to perceive how
key reservoir values change at each point over the data set.
Secondly is the classification. The objective is to distinguish
an arrangement of classes or lithofacies. The geoscientist at first
partitions the zone of enthusiasm into classes then, neural
networks will identify these classes throughout the seismic data
set. Probability or uncertainty estimations also provided by
classification which the sample-to-sample method unable to
provide. All in all, in both applications, the end results of the
analysis are an even more geographically significant outcome
than the first seismic inversion input.
2.3 THE NATURE AND CLASSIFICATION OF
CARBONATE POROSITY
In general, carbonate sediments have a more complex pore
system compared to siliciclastic sediments. This is generally
due to wide variation of grain shape. To characterize these
complex pore system, two main carbonate porosity
classification have been established. The first classification was
by Choquette and Pray [9], where they classified based on
genetic and geologically which was well suited for porosity
evolution studies important for exploration [1].
The second classification was by Lucia [10] based on the
earlier classification of Archie [11]. This classification is
mainly used in carbonate reservoir characterization. There are
two types of porosity namely, primary porosity and secondary
porosity. The primary porosity is related with the first
depositional surface of the sediments. That is, essential porosity
is the pore space in between the grains and inside the
depositional framework. This better known as intergranular
porosity. Pore space inside (or within) of fossil pieces or
openings within mineral grains that existed preceding to
deposition is also known as primary porosity, however is
mentioned as intragranular porosity [1].
Secondary porosity will be porosity that created after burial
and deposition in the sedimentary basin. This type of porosity
outcomes from dissolving of grains, depositional network, or
cement. Secondary intragranular porosity is the most
effectively noticeable secondary porosity. Secondary
intergranular porosity is sometimes hard to observe and
measure [1].
2.4 SEISMIC INVERSION
Seismic inversion has been used in many E&P field for
various purposes. The proper procedure is dictated by the
complexity of the geologic conditions, and the information
accessible. Possibility studies focused at a well area give us
understanding into which inversion technique will best
demonstrate a reliable division of the lithofacies and model the
reservoir thicknesses [11]. There are four types seismic
inversion namely, deterministic seismic inversion,
geostatistical seismic inversion, 4D seismic inversion and
3D/4D Petrophysical inversion.
2.5 SQp AND SQs
Gamma ray and resistivity are important reservoir
properties. Gamma ray log used to distinguish the sandstone
and the shale. Thus, it is using to predict the lithology.
Resistivity logs on the other hand give information about the
fluid content of the reservoir. Up until now, it is hard to acquire
the ideal pseudo logs of gamma ray and resistivity from elastic
properties that can be gotten from seismic data. Nevertheless,
various endeavors have been proposed to inexact the qualities
for lithology and fluid indicators. By performing rotation
technique into certain angles, one can approximate the
resistivity log and gamma ray log. This method known as
Amplitude Versus Offset (AVO) rotation technique under AVO
technology.
Timko [12] connected the seismic rotation technology to
deliver a pseudo gamma ray log to tackle the issue of sand/shale
interbedding in the lower Cretaceous of Near Base Cretaceous
(NBC). The forecast of resistivity from flexible properties has
likewise been rehearsed. Xu and Duta [13] utilized both P and
S waves to foresee resistivity, which can be gotten from seismic
inversion result. Mavko et al. [14] who utilized a model of plane
wave spread in the viscoelastic medium, as administered by the
Kramers– Kronig connection. By utilizing the Hudson Crack
Theory, the inverse Quality factor of the P and S waves (Qp and
Qs) were characterized as far as bulk and shear modulus. To
rationalize the reliance of the condition on constriction
parameter, the SQp and SQs qualities are created. According to
Hermana et al. [15], the SQp and SQs are formulated, as in:
(1)
where M/G is the ratio of bulk and shear modulus that can be
approximated from the P-wave and S-wave velocity ratio.
 METHODOLOGY Table 1: Summary of well log data for Well S-1 and S-105

3.1 RESEARCH AND DATA GATHERING

The microfacies determined throughout 10 outcrops is
Upon finalizing the title of the project, the next step was to
conduct a research and followed by data gathering. Research
was carried out based on the thesis papers, journals and books
that are related to rock physics studies, neural network and The zone of interests was determined for each of the wells
porosity prediction in carbonates. The papers reviewed are up- based on the given well log data and drilling reports. The table
to-date and highly relevant to the project. below (Table 2) summarizes the top of carbonate and zone of
interest for the two wells. The well log data and interpreted zone
3.2 RESEARCH WORK FLOW of interest of Well S-1 and S-105 can be seen in Figure 3 and 4
respectively.
There are few steps involved in the research work flow in
order to meet the expected results. Research work flow is Table 2: Summary of Top of carbonate and zone of interest
carried out using HampsonRussell and PETREL software. The
steps of research work flow as follows:
Seismic data
Well log data (near and far
stack)

Well log Seismic

analysis Inversion

Elastic Determine
properties elastic
calculation properties

Neural
Cross Plot Network
training

Estimate Predict Figure 3: Well log data display of Well S-1

Porosity porosity

Comparison
and checking

Figure 2: Research work flow of the project

3.3 ANALYSIS OF RESULTS

Analysis of the result was carried out from the output, which
was the porosity estimation from neural network and empirical
equation. The results were compared to actual porosity to
check for the accuracy of the porosity estimation in carbonate
rocks. The analysis of results was made based on the resource Figure 4: Well log data display of Well S-105
available. Well data will be used for validation of the results.
4.2 ELASTIC PROPERTIES CALCULATION

RESULTS AND DISCUSSIONS Few elastic properties such as P-impedance (AI), S-

4.1 WELL LOG ANALYSIS
The well log data of the wells in the study area only contains
certain data. The data given for each well as follows:
impedance (SI), Lambda Mu Rho (LMR), SQp and SQs were
calculated and generated for Well S-105 which were used to
create the cross-plots. The elastic property which able to
differentiate the porosity into high porosity to low porosity
range will be designated to be used as a training data for neural
network.
 Table 3: Formula used to calculate elastic properties
4.2.1 Micritization
The micritization features can be seen as light to dark brown
4.3 CROSS-PLOT ANALYSIS
The cross-plots were created between porosity versus
elastic properties. The main aim to create these cross-plots is to
check the efficiency of a certain elastic property as a parameter
to estimate porosity in carbonate reservoir. Besides, it is also
helpful to determine feasibility of the seismic inversion, before
proceeding to the full inversion workflow. The color coding
was assign based on porosity attribute to visualize better. The
results are as per below
Figure 4.15: Cross-plot of Porosity vs SQs
Figure 5: Cross-plot of Porosity vs SQs
Figure 6: Cross-plot of Porosity vs SQp
The SQs and SQp attributes tested to check the reliability to
predict porosity in carbonates. Figure 5 and 6 shows the cross-
plot between the porosity with SQs and SQp attribute
respectively. From the cross-plots, a trend of linear relation was
observed in both attributes. Porosity and SQs produced a good
correlation of 0.96662. Thus, this elastic property can be used
as an input to train the neural network to estimate porosity. On
the other hand, a poor correlation of 0.694145 was established
in SQp. Hence, this elastic property is not recommended to be
used as an input to train the neural network to estimate porosity
even though it can distinguish porosity into a higher and lower
value.
4.4 PREDICTED POROSITY FROM EMPIRICAL
EQUATION
SQs was selected as an input to estimate porosity from the
empirical equation. This is because the SQs yielded a good
correlation of 0.96662 with porosity. The equation used was as
below:
y = -1.75344x – 0.871635 (2)
where x was substituted by the original SQs equation, refer
Table 3.
A porosity model was created through the correlation of
SQs porosity (Figure 7). The predicted porosity from empirical
equation of SQs yielded a good correlation of 0.961911 with
the actual porosity from well data (Figure 8). It provided a
better and clearer indication of porosity in the carbonate body.
A higher amplitude can be noticed on the top of carbonate till
gas water contact and lower amplitude can be seen at the water
zone. The higher porosity is observed at the gas zone which
matches with the well log data. Thus, SQs attribute is highly
reliable to predict porosity in carbonate body.
4.2.2 Grain Alignment, Distortion Mechanical Grain
Packing, Grain Breakage & Grain Suturing
Figure 7: Porosity model from Qs porosity empirical equation
Figure 8: Cross-plot of Porosity vs Qs Porosity
 4.5 SEISMIC INVERSION
The author conducted pre-stack inversion on well S-1. Input
such as angle gathers, horizons, wavelets and initial model were
used to applied inversion process to the whole seismic volume.
As for the two horizons (Top of carbonate and Base of
carbonate) representing the carbonate platform were interpreted
by picking seismic amplitude events in all inlines available.
Initial models of P-impedance (Figure 9) was used to create
inverted elastic properties such as P-impedance, S-impedance,
density, Lambda-mu-rho (LMR), SQp and SQs. The result of
inverted models has an improved image with clearer and the
detailed layering allowed us to easily differentiate the lithology
and fluid present in our target carbonate reservoir.
4.2.3 Cementations (Pore Filling Cements)
The most common cements found are granular mosaic
Figure 9: Initial model for Pre-stack inversion
Figure 10: Inverted SQs
Figure 11: Inverted SQp
The SQs also showed the same pattern as density, P-
impedance and S-impedance but it is more detailed and have a
better visualization compared to the rest of the elastic
properties. The highest porosity value can be obtained at the top
of carbonate and at the flank up until GWC. The value relatively
decreases as it enters the water zone and the lowest value
recorded at the base of the carbonate. Additionally, the interval
from top of carbonate until GWC indicates an area with higher
porosity and permeability when compared to surrounding areas.
As a contrast, SQp showed the lowest value at the top of
carbonate until the GWC. The value started to increase as it
entered the water zone. The highest value recorded at the base
of carbonate.
The author decided to plot the porosity well log curves and
inserted variable color log of the elastic properties on the
models to check the inversion quality. Based on the porosity
curve, the porosity is higher in the gas zone compared to water
zone. The variable color log of the elastic properties showed
almost similar pattern as the model thus, the inversion is
reliable. There is an alternative way to check the inversion
quality, which is by cross-plotting the inverted model with
actual logs. From the cross-plot, errors in the SQs model
showed the lowest while the errors in the Zp model are the
highest (Figure 12).
The author could not compare the inverted models directly
with the geology of the Well S-1 of Field X. This is because,
there were no core samples and petrophysical data available in
Well S-1. Predominantly, the lack of petrophysical data led to
the use of data from analog rocks to estimate, for example, the
velocity of S-wave. Undoubtedly, the acquisition of S-wave
along this well should improve the inversion results.
Figure 12: Errors models
4.6 NEURAL NETWORK ANALYSIS
In this project, a model has been developed and employed
to check the accuracy and reliability of the porosity from new
seismic attribute in carbonate reservoir. A new parameter
namely SQs was considered as the input of the proposed
models; the porosity was the output.

Figure 13: Porosity model from Neural Network
Figure 14: Zoomed version of porosity model
The figure above shows the zoomed in version of computed
porosity from neural network. The author decided to plot the
porosity well log curves and inserted variable color log of the
porosity on the models to check the quality. The observed
porosity has roughly the same pattern as the computed porosity.
However, further analyzation is needed to check the accuracy
and its reliability. Thus, a cross plot was created (Figure 15).
S-1
Correlation = 0.891984
S-
105
Figure 15: Actual porosity vs predicted porosity
The actual porosity and predicted porosity produced a
good correlation of 0.891984. However, porosity estimated
from empirical equation produced slightly higher correlation of
0.961911 compared to the predicted porosity using neural
network.
CONCLUSION
The methodology adopted had been able to meet the
objectives of this project successfully. This project mainly
focuses on the application of new elastic properties namely SQp
and SQs in neural network to estimate the porosity and its
reliability in carbonate reservoir. Based on the results, it can be
concluded that both SQp and SQs is unfailing to predict
porosity in carbonate reservoir as it shows good trend in
distinguishing the higher and lower porosity values. However,
SQs was selected as an input for the empirical equation and
neural network. This is because, SQs shows better correlation
of 0.96662 than SQp. Besides, it’s also yielded less inversion
error which was about 0.035. Two porosity models were
created in this project. First model from the empirical equation
of SQs and the second model from neural network analysis with
inverted SQs as an input. Both the models were compared with
actual porosity from the well data to check its accuracy and
reliability. From the comparison, the porosity estimated from
empirical equation yielded slightly better correlation of
0.961911 compared to porosity estimated from neural network
of 0.891984. In conclusion, SQs is highly reliable to estimate
the porosity in carbonate reservoir through empirical equation
and neural network. Neural network highly endorsed to use in
the future as tool to estimate porosity in the carbonate reservoir.
FURTHER STUDY
As for further study, the author suggests conducting this
research on other carbonate reservoir to test the reliability of the
neural network and SQs on estimating porosity. Overall, it is
also recommended for expansion and continuation of this
project. This is mainly to determine the other petrophysical
properties such as permeability and water saturation to further
validate the estimation performance using neural network and
SQs in carbonate reservoir.
ACKNOWLEDGMENT
I would like to thank my supervisor, Dr. Maman Hermana
for the opportunity to conduct this study and his guidance and
support he provided throughout the course of my final year
project. Not to forget, my gratitude to Center for Subsurface
Seismic Imaging & Hydrocarbon Prediction (CSI), Universiti
Teknologi PETRONAS and PETRONAS for providing
proprietary data. Last but not least, I would like to thank all my
family and friends for their endless moral support.
REFERENCES
[1] C. H. Moore, "The Classification of Carbonate
Porosity," in Carbonate Reservoirs Porosity
 Evolution and Diagenesis in a Sequence Stratigraphic Attenuation," in Offshore Technology Conference
Framework, LA, ELSEVIER, 2001, pp. 37-59. Asia, Kuala Lumpur, Malaysia, 2016.
[2] M. A. R. D. &. L. R. A. Nikravesh, "Soft computing: [17] Greenberg, M.L and J.P. Castagna, "GreShear-wave
tools for intelligent reservoir characterization (IRESC) velocity estimation in porous rocks: Theoretical f
and optimum well placement (OWP).," Journal of ormulation, preliminary verification and applications:
Petroleum Science and Engineering, pp. 239-262, Geophysical Prospecting," Geophysical Propecting ,
2001. pp. 195-209, 1992.
[3] P. G. T. &. T. I. Wong, "An Imporved Technique in [18] Y. Egemen, "PETROPHYSICAL ANALYSIS AND
Porosity Prediction: A Neural Network Approach," ROCK-PHYSICS BASED PREDICTION OF SONIC
IEEE Transcations on Geoscience and Remote VELOCITIES IN CARBONATES," Michigan
Sensing, pp. 971-980, 1995. Technological University, Houghton, 2105.
[4] B. Russell, "Neural Networks find meaning in data,"
Janurary 2013. [Online]. Available: from
https://www.cgg.com/technicalDocuments/cggv_000
0016502.pdf.
[5] M. B. Madon , "Geological Setting of Sarawak," in
The Petroleum Geology and Resources of Malaysia,
Kuala Lumpur , PETRONAS, 1999, pp. 275 - 290.
[6] P. Tapponnier, G. Peltzer, A. Le Dain, R. Armijo and
P. Cobbold, "Propagating extrustion tectonics in Asia:
New insights from simple experiments with
plasticine," Geology, 1982.
[7] D. James, "The geology and hydrocarbon resources of
Negara Brunei Darussalam," in Regional geological
setting, Muzium Brunei, 1984, pp. 34-42.
[8] V. Vahrenkamp, I. Biling and J. Grotsch, "Carbon-
isotope stratigraphy in shallow-water carbonates:
implications for Cretaceous black-shale deposition,"
Sedimentology, pp. 623-634, 1998.
[9] P. Choquette and L.C.Pray, "Geologic nomenclature
and classification of porosity in sedimentary
carbonates," American Association of Petroleum
Geologists Bulletin, vol. 54, pp. 207-250, 1970.
[10] F. Lucia, "Petrophysical parameters estimated from
visual descriptions of carbonate rocks: a field
classification of carbonate pore space," Journal of
Petroleum Technology, vol. 35, pp. 629-637, 1983.
[11] G. Archie, "Classification of carbonate reservoir rocks
and petrophysical considerations," American
Association of Petroleum Geologists Bulletin, vol. 36,
pp. 278-298, 1952.
[12] "CGG," 10 08 2018. [Online]. Available:
https://www.cgg.com/en/What-We
Do/GeoConsulting/Geophysical-Services/Seismic-
Inversion.
[13] D. Timko, Pseudo Gamma Ray Volumes Estimated
from Offset Seismic, Canada: Recorder, 2003.
[14] Xu, H and Duta, N.C, "Prediction of Resistivity Using
P and S-wave Properties from Prestack Seismic
Inversion," in 68th EAGE Conference and Exhibition
incorporating SPE Europe, Vienna, Austria, 2006.
[15] Mavko, G, Mukerji, T and Dvorkin, J.P, The
Rockphysics Handbook, Cambridge, UK: Cambridge
University Press, 2009.
[16] Hermana, M, Ghosh, D.P and Sum, C.W, "Optimizing
the Lithology and Pore Fluid Separation Using