A hydrocarbon microtremor survey over a gas field: Identification of ...

A hydrocarbon microtremor survey over a gas field: Identification of seismic attributes 

Erik H. Saenger*, ETH Zurich and Spectraseis, Arnaud Torres, Spectraseis, Susanne Rentsch, FU Berlin, 

Marc Lambert and Stefan M. Schmalholz, ETH Zurich, 

Efrain Mendez-Hernandez, Pemex Exploracion y Produccion 

Summary 

Narrow-band, low-frequency (from ~1H to ~6Hz) 

microtremor signals have been observed worldwide at the 

surface over hydrocarbon reservoirs (oil, gas and water 

multiphase fluid systems in porous media). These lowfrequency 

ëhydrocarbon microtremorsí possess remarkably 

similar spectral and signal structure characteristics, 

pointing to a common source mechanism, even though the 

depth (some hundreds to several thousands of meters), 

specific fluid content (oil, gas, gas condensate of different 

compositions and combinations) and reservoir rock type 

(such as sandstone, carbonates, etc.) for each of those sites 

are quite different. In this extended abstract we describe 

some results of an extensive low-frequency hydrocarbon 

microtremor survey carried out in Mexico. We describe 

how we extract a hydrocarbon potential map from spectral 

anomalies and compare this map with available reservoir 

location data. In a detailed analysis of two representative 

stations of the survey we extract different seismic 

attributes. Those attributes are compared with a theoretical 

model which can explain the source mechanism of 

hydrocarbon microtremors 

Introduction 

A growing number of surveys at different oil and gas field 

locations throughout the world have established the 

presence of ëhydrocarbon microtremorsí with a high degree 

of correlation to the location and geometry of hydrocarbon 

reservoirs (Dangel et al., 2003; Holzner et al., 2005; Graf et 

al., 2007 and references therein). These tremors can be used 

as a direct hydrocarbon indicator for the optimization of 

borehole placement during exploration, appraisal and 

production. The ever-present seismic background noise of 

the earth (e.g., Berger et al., 2004) acts as the driving force 

for the generation of hydrocarbon indicating signals. In 

contrast to conventional 2D and 3D seismic technologies, 

the investigation of ëhydrocarbon microtremorsí is entirely 

passive and does not require artificial seismic excitation 

sources. 

In this paper we describe some results of a specific survey 

carried out by Spectraseis over a gas field in Mexico. By 

using ultra-sensitive, portable 3C broadband seismometers 

(sensor type: G¸ ralp) more than 500 measurements of the 

ever-present seismic wavefield at the surface were acquired 

over an area of approximately 200 km 2 . We will briefly 

SEG/San Antonio 2007 Annual Meeting 

describe the main processing steps and the resulting 

hydrocarbon potential map. 

In a second step we analyse the recorded data at two 

locations in detail. We extract several key seismic attributes 

to characterize the hydrocarbon microtremor signal. The 

goal of this characterization is twofold: On the one hand 

this will solidify or falsify theoretical explanations of the 

origin of hydrocarbon microtremors. On the other hand the 

knowledge of those attributes will improve the processing 

and interpretation for future surveys. 

The survey: Acquisition, Processing and Interpretation 

The selected survey area lies in the Burgos Basin area in 

the north-western part of Mexico. The origin of this basin is 

associated to the opening of the Gulf of Mexico, during the 

Jurassic, starting the sedimentation in the Callovian with 

evaporitic deposits. The sedimentation conditions changed 

in the Cenozoic when a great regression occurred and a 

sedimentary sequence of at least 8000 meters thickness, 

with a great gas potential (represented by rocks with Type 

III Kerogen) was accumulated. The complex fault system 

in the surveyed area is made up of horst and graben 

structures, and is part of a large half-graben system (fault 

style is mainly listric). Minor faults compartmentalize the 

whole system in smaller blocks. 

A sensor grid layout is used with node spacing ranging 

from 250 to 1000 m. Several monitoring stations are 

installed for the duration of the entire survey. After each of 

the measurements (duration from 4 hours to 24 hours) the 

raw 3C sensor data (surface velocities) was stored with an 

individual identification number (id). 

The raw data may include strong perturbations (noises, 

artefacts) and discontinuities (data gaps). In order to obtain 

a clean signal in the time domain we cut out those time 

intervals with obvious strong artificial signals. 

From the cleaned data we calculate the power spectral 

density (PSD). An applied standard procedure is to 

determine the PSD for 40 sec. time intervals and to 

calculate the arithmetic average of each PSD for the whole 

measurement time. This leads to a stable and reproducible 

result in the frequency domain. In Figure 1 we show two 

spectra for the vertical component. 

1277

Figure 1: Spectrum of the passive seismic wavefield 

(vertical surface velocities) in the frequency range from 0.5 

to 7.4Hz. The id 70139 was recorded over a known gas 

field, id 70575 is over an area with no hydrocarbon 

potential. Both locations are marked in Figure 3. 

One goal of the processing of hydrocarbon microtremor 

data is to map low-frequency energy anomalies in the 

expected total bandwidth of the hydrocarbon microtremor 

(i.e. between ~1 and 6Hz). We suggest to use a special 

integration 

technique: This 

method considers 

only the vertical 

component of the 

signal. The noise 

variations are taken 

into account by 

defining an 

individual minimum 

for each spectrum 

between 1 and 1.7Hz 

(for hydrocarbon 

microtremors we typically observe a minimum in this 

range). The integral above this level is used for determining 

the KTI-IZ value (see Figure 2; KTI is a project 

abbreviation an IZ stands for Integral of Z-component). For 

this Mexican gas survey we calculate the integral only 

between 1 and 3.7Hz because of some identified artificial 

noise sources above this frequency interval. As an output 

we receive a hydrocarbon potential map (Figure 3) based 

on this specific integral attribute. 

The strong KTI-IZ values are well aligned with the zone 

where most of the gas-producing wells are currently 

concentrated. This match is quite satisfactory since it shows 

that it was possible to detect the presence of hydrocarbons 

in the most exploited part of the field. 

Seismic attributes of hydrocarbon microtremors 

Figure 2: Green surface 

defines the KTI-IZ value 


Figure 3: Subset of a hydrocarbon potential map (3 levels 

of probability for the presence of gas ñ orange is the 

highest) together with the KTI-IZ method values 

(ì bubblesî showing integrated peak surfaces for Z 

component). Two location are marked with their 

identification number (id). 


Now we have a detailed look on the measurement-point 

already shown in Figure 1. The KTI-IZ value described 

above is one attribute to characterize hydrocarbon 

microtremor signals. Another signature can be extracted by 

analyzing spectral ratios. In contrast to the well known H/V 

ratio method to identify soil layers we concentrate our 

analysis on the V/H ratio (i.e. the opposite). As shown in 

Figure 4 one observes a peak (values significantly above 1) 

in this ratio in the frequency band of hydrocarbon 

microtremors (i.e. ca. 1..6Hz) for a station placed above 

hydrocarbons. For details how we calculate this ratio refer 

to Lambert et al. (2006). 

1278

Figure 4: V/H ratio of the passive seismic wavefield in the 

frequency range from 0.5 to 10Hz. The id 70139 was 

recorded over a known gas field (left hand side), id 70575 

is over an area with no hydrocarbon potential. The dashed 

line indicates a value of V/H=1. 

It is also possible to perform a polarization analysis of the 

recorded microtremor data. The first step in this procedure 

is to bandpass-filter the 3C-data in the time domain. We 

apply a zero-phase filter which cuts out the frequencies 

from 1Hz to 3.7Hz for further analysis. In this work, the 

analysis of the polarization behavior over 40sec.-time 

intervals was adapted from Jurkevics (1988). Considering 

any time interval of three-component data ux, uy and uz 

containing N time samples auto- and cross-variances can be 

obtained with: 

C 

⎡ 1 

⎢ 

⎣ N 

N 

= ∑ 

s= 


⎤ 

ui( 

s) 

u j ( s) 

⎥ 

⎦ 

ij 

1 

where i and j represent the component index x, y, z and s is 

the index variable for a time sample. The 3◊3 covariance 

matrix 

⎛C 

⎜ 

C = ⎜C 

⎜ 

⎝C 

xx 

xy 

xz 

C 

C 

C 

xy 

yy 

yz 

C 

C 

C 

xz 

yz 

zz 

⎞ 

⎟ 

⎟ 

⎟ 

⎠ 

is real and symmetric and represents a polarization ellipsoid 

with best fit to the data. The principal axis of this ellipsoid 

can be obtained by solving C for its eigenvalues λ1 7λ2 7λ3 

and eigenvectors p1, p2, p3: 

( C − λ I ) p = 0 

where I is the identity matrix. 

The parameter called rectilinearity L, sometimes also called 

linearity, relates the magnitudes of the intermediate and 

smallest eigenvalue to the largest eigenvalue 

⎛ λ2 

+ λ3 

⎞ 

L = 1− 

⎜ , 

2 ⎟ 

⎝ λ1 

⎠ 

and measures the degree of how linear the incoming 

wavefield is polarized. It yields values between zero and 

one. The other two polarization parameters describe the 


orientation of the largest eigenvector p1 = (p1(x), p1(y), 

p1(z)) in dip and azimuth. The dip can be calculated with 

⎛ 

φ = arctan⎜ 

⎜ 

⎝ 

p ( x) 

1 

p ( z) 

1 

2 

+ p ( y) 

and is zero for horizontal polarization and is defined 

positive in positive z-direction. The azimuth is specified as 

⎛ p ⎞ 1( 

y) 

θ = arctan ⎜ 

⎟ 

⎝ p1( 

x) 

⎠ 

and measured positive counterclockwise (ccw) from the 

positive x-axis. In addition we analyse the strength of the 

signal which is given by the eigenwert λ1. All four 

attributes are illustrated in Figure 5. 

Figure 5: Polarization parameter sketch. Reliability of dip 

and azimuth. Left hand side: high rectilinearity and 

medium dip, Right hand side: low rectilinearity and 

relatively high dip. The length of the red arrow is given by 

the largest eigenvalue λ1, further refered as the strength of 

the signal. 

In Figure 6 and 7 we apply the analysis explained above to 

our microtremor data of station id 70139 and 70575 with 

high and low hydrocarbon potential, respectively. It is clear 

that we have on the surface a complicated mixture of 

different wave types with different origins. However, we 

have identified very stable trends in the selected gas survey 

regarding the attributes dip, azimuth, rectilinearity and 

strength in the hydrocarbon microtremor frequency range 

(i.e. 1..3.7Hz). We summarize our observations as follows: 

Above a reservoir (station id 70139): 

- Dip: Stable high value (≥80 o ) directly above the 

reservoir (Figure 7, top, left hand side). 

- Strength: Varying, but present over the whole 

measure period. 

- Rectilinearity: Relative high and relatively stable 

and somehow correlated with the strength 

- Azimuth: Strongly varying, as expected for such 

high dip values. 

1 

2 

⎞ 

⎟ 

⎟ 

⎠ 

1279

For a measure point with low hydrocarbon potential 

(station id 70575): 

- Dip: Stable low value (≈20 o , Figure 7, top, left hand 

side). 

- Strength: Relatively low with some spikes. 

- Rectilinearity: Lower in comparison with the values 

observed above a hydrocarbon reservoir. 

- Azimuth: Relatively stable; maybe points to an 

artificial noise source. 

Figure 6: Time variations of dip (φ), strength (λ1), 

rectilinearity (L) and azimuth (θ) for bandpass-filtered 

data (1..3.7Hz) from station id 70139 (High hydrocarbon 

potential). The time unit on the horizontal axes is periods of 

40sec. The solid line represents the value using data of the 

whole time period. 

Figure 7: Same as Figure 6 but for station id 70575 (Low 

Hydrocarbon potential) 

A theoretical model based on observed attributes 

In Graf et al. (2007) we discuss some possible origins of 

hydrocarbon microtremors. Based on the seismic attributes 

identified in this paper we suggest the following theoretical 

model. The driven sources are ocean waves interacting with 

the coast structure. They produce the so-called ocean-wave 



peak around 0.1Hz in the seismic background noise (e.g. 

Berger et al. 2004) which can be observed on all locations 

around the world. Those surface waves propagate through 

whole continents and can be used for determining seismic 

velocities down to a depth of 20km (e.g. Shapiro et al. 

2005). Interestingly, Rayleigh waves around 0.1Hz 

oscillate at reservoir depth (deeper than ≈500m) mainly in 

vertical direction (e.g. Aki and Richards 2002). By a 

complex resonant amplification effect (see Graf et al. 2007 

and references therein) with the hydrocarbon reservoir the 

microtremor itself produce a radiation pattern as shown in 

Figure 8. Directly above the reservoir one observe mainly 

P-waves. This is consistent with the following observations 

at id 70139 of the selected gas reservoir survey: Strong 

anomaly in the vertical component (Figure 1), a peak in the 

V/H-ratio (Figure 4), a constant high dip with a relatively 

high rectilinearity and an non-vanishing strength as shown 

in Figure 6. 

Figure 8: A model which explains the observed seismic 

attributes. 

Conclusions 

We have described a low-frequency hydrocarbon 

microtremor survey over a gas field in Mexico. As in 

previous case studies (Graf et al. 2007 and references 

therein) we observe a high correlation of the hydrocarbon 

microtremor signal with the known location of the gas 

reservoir. With a deeper analysis we have indentified and 

specified different seismic attributes of the signal: KTI-IZ 

value, V/H-signal, dip, rectilinearity and strength. Based on 

those attributes we propose a model of the origin of the 

described LF-signal. 

Acknowledgements 

We are gratefully to Pemex and Spectraseis for the 

permission to publish the data. Swiss KTI-program has cofinanced 

this work 

1280

EDITED REFERENCES 

Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2007 

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 

Aki, K., and Richards, P. G., 2002, Quantitative seismology, 2nd ed: University Science Books. 

Berger, J., P. Davis, and G. Ekstrom, 2004, Ambient earth noise: A survey of the global seismograph network: Journal 

Geophysical Research, 109, B11307; http://dx.doi.org/10.1029/2004JB003408. 

Dangel, S., M. E. Schaepman, E. P.Stoll, R. Carniel, O. Barzandji, E. D. Rode, and J. M. Singer, 2003, Phenomenology of 

tremor-like signals observed over hydrocarbon reservoirs: Journal of Volcanology and Geothermal Research 128, 135– 

158. 

Graf, R., S. M. Schmalholz, Y. Podladchikov, and E. H. Saenger, 2007, Passive low frequency spectral analysis: Exploring a new 

field in geophysics: World Oil, 228, 47–52. 

Holzner, R., P. Eschle, H. Zuercher, M. Lambert, R. Graf, S. Dangel, and P. F. Meier, 2005, Applying microtremor analysis to 

identify hydrocarbon reservoirs: First Break, 23, 41–49. 

Jurkevics, A., 1988, Polarization analysis of three-component array data: Bulletin of the Seismological Society of America, 78, 

1725–1743. 

Lambert, M., B. Steiner, S. M. Schmalholz, R. Holzner, and E. H. Saenger, 2006, Soft soil amplification of ambient seismic noise 

– field measurements and numerical modeling of H/V ratios: Passive Seismic Workshop, EAGE, A10. 

Shapiro, N. M., M. Campillo, L.Stehly, and M. Ritzwoller, 2005, High resolution surface wave tomography from ambient 

seismic noise: Science, 307, 1615–1618. 


1281

A hydrocarbon microtremor survey over a gas field: Identification of ...

Create successful ePaper yourself

Delete template?

Save as template?