GEOLOGICA CARPATHICA, JUNE 2005, 56, 3, 285294
www.geologicacarpathica.sk
Seismic reflection investigations for gas storage in aquifers
(Mura Depression, NE Slovenia)
ANDREJ GOSAR
1, 2
1
Environmental Agency of the Republic of Slovenia, Dunajska 47, SI-1000 Ljubljana, Slovenia; andrej.gosar@gov.si
2
University of Ljubljana, Faculty of Natural Sciences and Engineering, Askerèeva 12, SI-1000 Ljubljana, Slovenia
(Manuscript received January 7, 2004; accepted in revised form June 16, 2004)
Abstract: Two antiform structures in the Mura Depression were investigated as the most promising in Slovenia for the
construction of an underground gas storage facility in aquifers. Seventeen reflection profiles with a total length of
157 km were recorded, and three boreholes were drilled at their locations. Structural models based on interpretation of
the reflection seismic data, were constructed for the two main horizons (the pre-Tertiary basement and the Badenian/
Sarmatian boundary). Evaluation of different seismic velocity data was carried out to establish lateral velocity variations
in order to perform correct time-to-depth conversion. The porous rock in the Pecarovci structure is a 70 m thick layer of
dolomite, occurring at a depth of 1900 m, whereas layers of marl, several hundred meters thick, represent the imperme-
able cap rock. Due to faults, the Dankovci structure, at a depth of 1200 m where the reservoir rocks consist of thin layers
of conglomerate and sandstone, was found to be less reliable. 1D synthetic seismograms were used to correlate the
geological and seismic data at the borehole locations. Raytracing modelling was applied to confirm lateral continuity of
some horizons and to improve the structural interpretation at the locations of faults, which are critical factors for the
storage of gas.
Key words: Pannonian Basin, Mura Depression, raytracing modelling, synthetic seismograms, gas storage, seismic
reflection.
Introduction
Slovenia imports almost all the gas it needs. The supply
through the pipelines is fairly constant throughout the year,
but consumption is subject to seasonal changes. For this rea-
son, gas must be stored in the summer months in order to per-
mit higher consumption during the winter. For economic and
safety reasons, the storage of natural gas is reasonable only
when underground (Dussaud 1989). There are four main types
of underground storage: salt caverns, abandoned mines, aqui-
fers or depleted oil or gas fields, and hard rock caverns (Gaus-
sens 1986).
The geological structure of Slovenia only permits storage in
aquifers (Fig. 1). Geological investigations for such facilities
have been ongoing for more than 15 years. The goal has been
to find an appropriate antiform structure, at a depth of between
500 and 2000 m, composed of porous (reservoir) rock with an
impermeable covering layer (Tek 1989; Flanigan 1995). In the
first stage, 13 different locations (Sadnikar 1993) were investi-
gated and two of them (Pecarovci and Dankovci) in the Mura
Depression were selected for further exploration. Geophysical
methods, especially reflection seismics, had an important role
in the evaluation of possible suitable locations.
The Pecarovci and Dankovci structures are located in the
Mura Depression (Gosar 2005), on the slope of the Murska
Sobota massif, which dips towards the Radgona depression. In
this area, the depth to the pre-Tertiary basement (Pt horizon) is
between 1800 and 2000 m. The possible collector rocks for
gas storage are the Mesozoic carbonates in the pre-Tertiary
basement and the thin layers of porous conglomerates and
sandstones above the discordant boundary between the Bade-
nian and Sarmatian layers (KB horizon) inside the Tertiary
sediments (Turk 1993).
Seismic reflection investigations
A dense net of reflection seismic profiles was recorded at
the Pecarovci-Dankovci location. For the construction of a
structural model in an area measuring 8×8 km (Fig. 2), we
used 17 profiles. The total length of these profiles is 157 km
and their length inside the modelled area is 94 km. Geofizika
Zagreb performed the field data acquisition. Most of the pro-
files were recorded with line explosive sources (Geoflex and
Fig. 1. Schematic drawing of gas storage facility in an aquifer (af-
ter Gaussens 1986).
286 GOSAR
Primacord), end-offset layout and a Texas Instruments DSF
IV seismograph. The distance between geophone groups was
40 m, and CMP coverage was 24- or 30-fold. Three profiles
(Pec-1v-89, Dan-1v-89 and Pec-Dan-1v-89) were recorded
with Vibrosise source, split offset layout, a Texas Instruments
DSF V seismograph, 30 m geophones group spacing and 24-
fold coverage. Linear geophone arrays composed of 24 SM-4/
UB 10 Hz geophones were used. Standard data processing
Fig. 2. Location map of the seismic profiles and boreholes at the PecarovciDankovci location.
was performed by INA-Naftaplin in Zagreb and by Western
Geophysical in London.
In the region of the Pecarovci and Dankovci structures,
three deep boreholes have been drilled to date (Fig. 2). Two of
them (Dan-1 and Pec-1) have reached the pre-Tertiary bedrock
at a depth of between 1800 and 1950 m, while Dan-3 has ter-
minated below the KB horizon at a depth of 1400 m (Sadnikar
1993). In the older Dan-1 borehole drilled for oil exploration,
SEISMIC REFLECTION INVESTIGATIONS FOR GAS STORAGE (SLOVENIA) 287
only basic electrical well-logging was performed. In the Dan-3
and Pec-1 boreholes, a complete set of well-logging measure-
ments was performed, including sonic velocity measurements.
In the Dan-3 and Pec-1 boreholes, down-hole seismic velocity
measurements were also performed. In the Mesozoic dolo-
mites, overlying metamorphic complex thermal water with the
temperature of 102 °C was drilled in the Pec-1 borehole at a
depth of 1850 m (Sadnikar 1993). This borehole (Fig. 3) has
drilled the Mura, Lendava and Murska Sobota Formations
(Gosar 2005). The main horizons interpreted in all seismic re-
flection profiles in the region are the KB horizon, which corre-
sponds to the discordant boundary between the Badenian and
Sarmatian sediments, and the Pt horizon, which corresponds
to the pre-Tertiary carbonate or metamorphic bedrock.
1D synthetic seismograms
Seismic modelling in one dimension (1D) is a commonly
applied tool for the correlation of seismic (time) data with geo-
logical or well-logging (depth) data (Sheriff & Geldart 1995).
This is important because, for the seismic data, lower vertical
resolution limited by the wavelength of the signal is character-
istic, but good lateral coverage along the profile is also appar-
ent. On the other hand, data from boreholes has good vertical
resolution, but is limited in lateral extent (Neidell 1981). In the
case of 1D modelling, the geological structure near the bore-
hole is approximated by horizontal layers. If the wavelength
Fig. 3. Lithologic column, interval velocity, density, acoustic impedance, reflection coefficient and two way time diagrams for the
Pec-1 borehole.
of the signal is small compared to the distance between adja-
cent interfaces, then good separated reflections are obtained.
But, in cases where the layers are thin with respect to the
wavelength, the seismic trace is a result of the interference of
signals reflected from several interfaces. The theoretical verti-
cal resolution of seismic data is 1/4 of the signal wavelength,
whereas in practice it is not greater than 3/8 of the wavelength
(Widess 1973).
At the PecarovciDankovci location, very thin layers and
the problem of interference close to the KB horizon were en-
countered. 1D modelling was therefore applied in order to im-
prove the interpretation of the KB horizon, and to correlate the
seismic and borehole data. For this modelling, synthetic seis-
mograms were constructed for boreholes Pec-1 and Dan-3 us-
ing the Vista software package (Sis 1990). The reflectivity se-
ries was computed from the sonic and density log data. The
sonic log was corrected on the basis of the down-hole mea-
surements. In the lowest part, close to the Pt horizon, the re-
sults of laboratory measurements of velocity on cores were
also used. In the Pec-1 borehole, 49 layers of different acous-
tic impedance were distinguished (Fig. 3). The highest reflec-
tion coefficients corresponded to the top and bottom of the
thin conglomerate layers above the KB horizon. We compared
the synthetic seismograms with two seismic profiles measured
close to the borehole location. Comparing the synthetic traces
with the Pec-Dan-1v-89 profile (Fig. 4b), which is trending
SWNE, good correlation can be observed at 0.5 s, near the
KB horizon (0.9 s), and at the Pt horizon (1.35 s), but poor
288 GOSAR
correlation between 0.7 and 0.9 s. If the same synthetic traces
are compared with the Pec-1v-89 profile, which is trending S
N (Fig. 4a), there is also a good correlation at both main hori-
zons and at 0.5 s, but two or three good reflections between
0.7 and 0.85 s not visible on the previous profile can also be
observed. It can be concluded that, in this area, a high degree
of velocity anisotropy is encountered. 1D modelling was also
used to test what influence the frequency content of the input
signal has on the vertical resolution of the seismic data to sup-
port advanced processing of the data (Yilmaz 1987).
Raytracing modelling
With 2D raytracing modelling stacked seismic sections,
field shot records and CMP (common midpoint) gathers were
simulated (Fagin 1991). 2D modelling was performed using
the Sierra Quik package (Sierra 1990) on the models con-
structed with the Sierra Mimic program. Quik programs use
the asymptotic raytracing theory methods to find the path of
seismic energy between the source and receivers (May & Cov-
Fig. 4. Comparison of the synthetic seismogram for the Pec-1 borehole with the profiles Pec-1v-89 (a) and Pec-Dan-1v-89 (b).
ey 1981). At each intersection of the ray with a horizon, the
program computes the time and the reflection coefficient. In
each layer, the program uses straight rays even if the velocity
varies. This approximation is good enough for most models.
At interfaces, the rays are refracted according to Snells law.
In layers with variable velocity, the direction of the ray is cal-
culated in following manner. When the ray enters the layer,
the local velocity is used to determine direction. The ray then
continues straight to the next interface, where a new local ve-
locity is used to compute the direction in the lower layer. The
result of a simulation is a spike section of reflection coeffi-
cients. The convolution of a spike section with the input wave-
let results in a synthetic seismogram. Using this method, fully
complex seismic amplitudes can be obtained in geological
structures of almost arbitrary configuration.
By 2D modelling of stacked seismic sections, an attempt
was made to confirm the continuity of some reflections related
to the KB horizon and the interpretation of faults at the Pt ho-
rizon (Gosar 1995). The results of the simulation are presented
for the characteristic profile, Pec-Dan-1v-89 (Fig. 5). The ba-
sis for interpretation and construction of the time model was
SEISMIC REFLECTION INVESTIGATIONS FOR GAS STORAGE (SLOVENIA) 289
Fig. 5. The unmigrated (a), and migrated (b) seismic profile Pec-Dan-1v-89.
290 GOSAR
the migrated seismic section (Figs. 5b and 6). When evaluat-
ing the structural interpretation of the Pt horizon, it was
proved that better results are obtained with modelling of the
unmigrated seismic section (Fig. 5a), where faults are more
Fig. 6. Line drawing interpretation of the seismic profile Pec-Dan-1v-89.
Fig. 7. Normal incidence raytracing for the profile Pec-Dan-1v-89.
evident because of diffractions. The input model consists of
nine layers of different velocity (Figs. 6 and 7). The two thin
layers of higher velocity represent the conglomerate sequenc-
es above the KB horizon. Between the Pecarovci and Dankov-
SEISMIC REFLECTION INVESTIGATIONS FOR GAS STORAGE (SLOVENIA) 291
Fig. 8. Synthetic seismograms for normal-incident raytracing shown in Fig. 7 (a) and for the corrected model (b).
ci structures there is a fault at which one sequence terminates
and the second is displaced. Above the Pt horizon, there is a
50 m thick layer of breccia. Normal incidence raytracing,
which simulates the unmigrated seismic section, is shown on
the depth model of this profile in Fig. 7 and the corresponding
synthetic seismogram in Fig. 8a. By comparing the synthetic
and the original seismic section (Fig. 5a), it was concluded
that the structure on the NE side of the Pecarovci antiform is
more complex. To prove this, a new model of the Pt horizon
was constructed at this location with two normal faults on the
NE side of the Pecarovci antiform instead of only one. This
model was simplified to only one interface because it was rec-
ognized that the upper layers do not affect the rays significant-
ly. The synthetic seismogram for this simulation (Fig. 8b)
showed better correspondence with the seismic section
(Fig. 5a). Therefore, we concluded that the Pecarovci structure
has, on its NE side, at least two normal faults. It is possible
also that the structure is even more complex.
Structural models
For structural modelling based on seismic reflection data, a
square area, 8×8 km in size, was selected. Seventeen profiles
with a total length of 94 km, and data from three boreholes,
were used (Fig. 2). The aim of the structural modelling was to
construct time and depth maps of the two most important hori-
zons, that is the KB the top boundary of the Badenian
292 GOSAR
rocks, and the Pt pre-Tertiary basement. Computer assisted
contouring of interpreted surfaces, taking into account fault
traces, was applied.
First, a detailed analysis of the available velocity data was
carried out to enable correct time-to-depth conversion. Four
types of velocity data were used: a velocity analysis from
seismic data; b down-hole measurements in boreholes; c
sonic logs; and, d laboratory measurements on cores. The
velocity function was based on the sonic log data, which were
corrected using the down-hole measurements, and fitted to the
datum plane of the seismic profiles (150 m above sea level).
Laboratory measurements on cores were used in the deeper
part of the Pec-1 borehole, where no other data was available.
Reflection velocity analysis data was used mainly to deter-
mine lateral velocity changes. The velocity isolines for the
characteristic profile Pec-Dan-1v-89 measured in the SWNE
direction showed no significant lateral velocity variations in
the upper part of the section, until a two-way time 1.0 s was
reached. However, there was a slight increase in velocity in
the NE direction, between 1.0 and 1.5 s. Established velocity
variation was applied for time-to-depth conversion.
The structural model of two main horizons (KB and Pt),
showing two-way traveltime isochrons of reflected waves,
was first constructed (Figs. 9 and 10). A structural time map of
the KB horizon (Fig. 9) shows a closed antiform structure at
Dankovci confined by the 980 ms (milliseconds) isochron,
which has a peak at 950 ms. Five faults cut the structure, but
they do not indicate significant slips. The reservoir rocks in
this horizon are thin layers of conglomerate and sandstone.
Small quantities of oil and gas were found in these layers,
proving the tightness of the cap rocks. On the other hand, be-
cause of the thin layers (a couple of meters thick), a fault could
easily separate two parts of the layer and reduce the volume of
the reservoir. At the Pecarovci location, there is no closed an-
tiform structure in the KB horizon.
On the Pt horizon (Fig. 10), there are closed antiform struc-
tures at both locations. The reservoir rock is a layer of porous
dolomite, approximately 70 m thick, overlying metamorphic
rocks in the basement. At Dankovci, it is confined by the
1460 ms isochron, while the top of the structure is at 1350 ms.
The area of the closed part is 5.42 km
2
. At Pecarovci, the top
of the structure is at 1350 ms and the antiform is confined by
the 1400 ms isochron. The highest point of the opening is on
the SW side. The area of the closed part is 1.576 km
2
. The pre-
liminary interpretation with a single fault on the NW side of
the anticline is presented in the model in Fig. 10. According to
raytracing modelling, the new interpretation with two normal
faults is shown in Fig. 11, together with a prognostic geologi-
cal profile. A minor reverse fault is also seen on the SW side
of the structure.
On the basis of the time-to-depth conversion, structural
depth maps were constructed for both horizons. The depth
from the datum plane to the KB horizon is from 1100 m to
1200 m, and the depth to the Pt horizon is from 1900 m to
2000 m (Fig. 11b)
Of the three antiform structures, the Pecarovci (Pt) structure
was selected for further investigations as the most promising
for the construction of the gas storage facility. The structure at
Dankovci (Pt) is too big for the desired storage volume, while
the structure at Dankovci (KB) was found to be less reliable
because of faults (Sadnikar 1993).
Conclusions
On the basis of the constructed models, the available storage
volume in the Pecarovci (Pt) structure (Fig. 11) was estimated.
The calculations were performed using the Evasit Program for
the evaluation of porous gas storage facilities (Gaz de France
1990). The input data comprised the areas of closed isoch-
Fig. 9. Time structural map of the KB horizon, equidistance = 20 ms,
x digitized points.
Fig. 10. Time structural map of the Pt horizon, equidistance = 20 ms,
x digitized points.
SEISMIC REFLECTION INVESTIGATIONS FOR GAS STORAGE (SLOVENIA) 293
Fig. 11. Planned underground gas storage facility in the Pecarovci (Pt) structure: a time structural map, b depth structural map,
c prognostic geological profile AA.
rones and corresponding seismic velocities in the top rock. As
the working volume, fifty percent of the total volume was tak-
en. Using this data, the working volume in the Pecarovci (Pt)
structure was estimated on 275300 million m
3
(n). This is
above the desired minimum volume of 200 million m
3
(n) se-
lected for the project of gas storage in Slovenia (Sadnikar
1993).
Among all evaluated locations in Slovenia, the antiform
structure in the pre-Tertiary basement at Pecarovci was proved
to be the most promising for the construction of an under-
ground gas storage facility. Its structure is defined by seven
seismic profiles and one borehole. To prove the structural in-
terpretation of seismic data and to test the hydrogeological pa-
rameters of the reservoir layer and the impermeable cap rock,
another four boreholes are planned. The main disadvantage of
this structure is the great depth of the storage layer, which re-
quires compressors of higher power, and higher costs during
operation.
Acknowledgments: The author is grateful to M. Bielik, F.
Hubatka and J. efara for constructive reviews of the article.
Part of this work was conducted while the author was receiv-
294 GOSAR
ing a grant from the University of Trieste. The assistance and
many valuable suggestions of R. Nicolich are gratefully ac-
knowledged. The author is indebted also to D. Ravnik, J. Sad-
nikar and S. Kranjc for their help and constructive remarks.
References
Dussaud M. 1989: Review of World Wide Storage Projects-France.
In: Tek M.R. (Ed.): Underground Storage of Natural Gas. Klu-
wer Academic Publishers, Dordrecht, 2329.
Fagin S.W. 1991: Seismic modelling of geological structures. Soci-
ety of Exploration Geophysicists, Tulsa, 1288.
Flanigan O. 1995: Underground gas storage facilities, design and
implementation. Gulf Publishing, Huston, 1179.
Gaussens P. 1986: Stockages souterrains de gas, Titre XIII. In: Man-
uel pour le transport et la distribution du gaz. Association tech-
nique de lindustrie du gaz en France, Paris, 1333.
Gaz de France 1990: Evasit, PC Version, Users Manual. Gaz de
France, Paris, 125.
Gosar A. 1995: Modelling of seismic reflection data for under-
ground gas storage in the Pecarovci and Dankovci structures-
Mura depression. Geologija 3738, 483549 (in Slovenian).
Gosar A. 2005: Geophysical and structural characteristics of the pre-
Tertiary basement of the Mura Depression (SW Pannonian Ba-
sin, NE Slovenia). Geol. Carpathica 56, 2, 103112.
May B.T. & Covey J.D. 1981: An inverse ray method for computing
geological structures from seismic reflections: zero offset case.
Geophysics 46, 4, 26887.
Neidell N.S. 1981: Stratigraphic modelling and interpretation: geo-
physical principles and techniques. Education Course Note Se-
ries 13. AAPG, Tulsa, 1141.
Sadnikar J. 1993: Investigations for underground gas storage in Slo-
venia. Rud.-Metalur. Zbor. 49, 12, 149167 (in Slovenian).
Sheriff R.E. & Geldart L.P. 1995: Exploration seismology. Cam-
bridge University Press, Cambridge, 1592.
Sierra 1990: Quik raytracing, Release 3.2, User Manual. Sierra Geo-
physics, Seattle, 1497.
SIS 1990: Vista 6.5, User Manual. Seismic Image Software, Calgary,
1134.
Tek M.R. 1989: Underground storage of natural gas. Kluwer Aca-
demic Publishers, Dordrecht, 1458.
Turk V. 1993: Reinterpretation of chronostratigraphic and lithos-
tratigraphic correlations in the Mura depression. Rud.-Metalur.
Zbor. 49, 12, 149167 (in Slovenian).
Widess M.B. 1973: How thin is a thin bed? Geophysics 38, 6,
11761180.
Yilmaz O. 1987: Seismic data processing. Society of Exploration
Geophysicists, Tulsa, 1526.