Professional Documents
Culture Documents
Ida Öhman
Eero Heikkinen
Tomas Lehtimäki
December 2006
POSIVA OY
FI-27160 OLKILUOTO, FINLAND
Tel +358-2-8372 31
Fax +358-2-8372 3709
Working Report 2006-114
Ida Öhman
Eero Heikkinen
Tomas Lehtimäki
Pöyry Environment Oy
December 2006
ABSTRACT
Posiva Oy takes care of the final disposal of spent nuclear fuel in Finland. In year 2001
Olkiluoto was selected for the site of final disposal. Currently construction of the
underground research facility, ONKALO, is going on at the Olkiluoto site.
The aim of this work was to use two-dimensional reflection seismic processing methods
to refraction seismic data collected from the ONKALO area in year 2002, and to locate
gently dipping reflectors from the stacked sections.
Processing was done using mainly open source software Seismic Unix. After the
processing, the most distinct two-dimensional reflectors were picked from seismic
sections using visualization environment OpendTect. After picking the features from
crossing lines were combined into three-dimensional surfaces. Special attention was
given for the detection of possible faults and discontinuities. The surfaces were given
coordinates and their orientation was adjusted using a geometric procedure, which
corresponds roughly a 3D migration, transferred to 3D presentation utility and
compared to available geological information.
The applied processing method was successful in detecting the reflectors. Most
significant steps were the refraction and residual statics, and deconvolution. Some
distinct reflectors can be seen at times 20-200 ms (vertical depths 50...500 m). The
signal gets noisier below 200 ms. Reflectors are best visible as coherent phase between
the adjacent traces, but do not raise much above the surrounding noise level. Higher
amount of traces to be stacked would emphasis the reflections and their continuity
more.
Reflectors picked on crossing lines match well to borehole observations (KR4, KR7,
KR24 and KR38) of fracture zones, and get support from geological and hydrological
models of the site. The observed reflections coincide with fracturing intensity and P-
wave velocity minima from boreholes. Reflections coincide also rather well to the
separate 3D seismic results from overlapping area.
The results demonstrate that seismic measurements intended for refraction interpretation
can also be successfully processed using reflection seismic processing methods.
Increasing number of active geophones and shots, and line density, would enhance
reliability of the reflections.
TIIVISTELMÄ
TABLE OF CONTENTS
ABSTRACT
TIIVISTELMÄ
1 INTRODUCTION .............................................................................................. 3
2 SOURCE DATA ................................................................................................ 5
2.1 Surface seismic data............................................................................. 5
2.2 Background information ........................................................................ 8
3 DATA PROCESSING ....................................................................................... 9
4 INTERPRETATION......................................................................................... 29
4.1 Two dimensional reflectors ................................................................. 29
4.2 Three dimensional surfaces ................................................................ 30
4.3 Comparison to borehole data.............................................................. 33
5 CONCLUSIONS.............................................................................................. 41
REFERENCES ........................................................................................................... 43
APPENDIX 1. STACKED SEISMIC SECTIONS AND PICKED REFLECTORS. ....... 45
2
3
1 INTRODUCTION
Posiva Oy takes care of the final disposal of spent nuclear fuel in Finland. In year 2001
Olkiluoto was selected for the site of final disposal. Currently construction of an
underground research facility, ONKALO, is going on at the Olkiluoto site. Suomen
Malmi Oy has carried out shallow refraction seismic surveys in 2001 and 2002 in order
to determine the overburden thickness and to study seismic P-velocity on bedrock
surface and to locate e.g. possible fracture zones (Ihalainen 2003). In year 2001 survey
lines from S1 to S27 were measured (15,4 km in total) and in year 2002 survey lines
from S27 to S69 (17,57 km in total).
The aim of this work was to use two-dimensional reflection seismic processing
methods to refraction seismic data collected from the ONKALO area, and to locate
gently dipping reflectors from the processed survey lines.
Processing applied open source software Seismic Unix (SU) (Stockwell & Cohen
2002). Processing included reading and re-arranging the data set, removal of noisy
traces, geometric spreading correction, bandpass filtering, refraction and residual
statics, deconvolution, airwave muting, automatic gain control, normal moveout and
stacking. The goal was to find the best possible processing parameters for the seismic
data set. The refraction measurement array is not optimized for reflection processing
which leads to quite small 2D fold, approximately only 4, which makes the processing
a demanding task.
2 SOURCE DATA
Original refraction seismic data was collected from Olkiluoto during the summer and
autumn 2002 (Ihalainen 2003). Survey was carried out by Suomen Malmi Oy (Smoy).
Data consists of 17 570 m of refraction seismic survey lines named S28-69. Lines S1 –
S27 of 2001 were recorded to 102,4 ms and were not considered in this work.
Original aim for the refraction field survey has been to determine the seismic velocities
in the bedrock surface, using refraction intercept time and generalised reciprocal
(GRM) interpretation techniques on 2D lines. The interpretations were further
examined and merged onto maps of soil thickness and velocity (Lehtimäki 2003).
Table 1 displays the information about the survey lines selected for the seismic
processing. The lines have been surveyed parallel on two directions (E-W and N-S)
and mainly at 100 m line spacing. Location of the lines is shown in Figure 1. Lines for
the seismic processing were selected to get the best possible coverage of the ONKALO
area within the available time schedule.
Table 1. Refraction seismic survey lines selected for the seismic processing.
Coordinates are locations for the first and last common mid points and length distance
between these points.
Line name Northing start Easting start Northing end Easting end Length m
28 6791770 1525650 6792010 1525650 240
282 6791995 1525650 6792230 1525650 235
29 6791727,5 1525700,00 6791980 1525700 252,5
30 6791800 1525750 6791950 1525750 150
302 6791920 1525750 6792250 1525750 330
31 6791770 1525800 6791920 1525800 150
312 6791900 1525800 6792230 1525800 330
32 6791740 1525850 6791890 1525850 150
322 6791890 1525850 6792232,5 1525850 342,5
33 6791830 1525950 6792250 1525950 420
34 6791795 1526050 6792137,5 1526050 342,5
342 6792047,75 1526050 6792230 1526050 182,25
35 6791805 1526150 6792292,5 1526150 487,5
362 6791960 1526250 6792192,5 1526250 232,5
372 6792030 1526300 6792170 1526300 140
43 6791950 1525490 6791950 1525730 240
432 6791950 1525720 6791950 1526230 510
44 6792000 1525500 6792000 1525650 150
442 6792000 1525655 6792000 1525790 135
443 6792000 1526130 6792000 1526280 150
45 6792050 1525547,5 6792050 1526330 782,5
46 6792100 1525480 6792100 1525900 420
47 6792150 1525340 6792150 1526390 1050
48 6792250 1526030 6792250 1526170 140
50 6792381,32 1526028,9 6792146,76 1526261,02 234,56
6
Length of each 24-channel geophone array was 90 meters (Figure 2). The geophone
spacing was 5 meters but at the both ends and in the middle of an array it was
supplemented to 2,5 meters.
Dynamite (35%) charges of 15-120 g were used as seismic source. The number of
shots for each array was nine, of which four were in-line offset shots 30-120 meters
outside the array and the rest at the both ends and in the middle of an array. Distances
between the shotpoints varied from 20 to 30 meters.
Line number 342 was surveyed using one meter geophone spacing and eleven
shotpoints on each 24 m long array.
Recording was done using type 10B, 4,5 MHz vertical geophones with 374 Ohm reel
manufactured by Mark Product’s (Houston, USA) (Ihalainen 2003). Acquisition was
carried out with ABEM Terraloc Mk VI seismograph. Sampling interval was 25 µs.
More detailed description of the field layout and equipment can be found in the
references Ihalainen (2003).
7
Figure 2. Survey array. At the line 342 geophone spacing was 1 meter and the number
of shots 11.
Seismic data was recorded to 409,6 ms (Figure 3 and Figure 4). Quality of the stacked
data is quite good down to 200 ms but after that it gets noisier. Quality of the data also
varies substantially from line to line as can be seen from Figure 3 and Figure 4. For
example lines S33 and S34 are located over a rocky outcrop and due to that the data is
quite noisy and therefore harder to process with uniform work flow settings. As can be
seen in Figure 3, in good data for example first arrivals (P-wave), S-wave and airwave
are easily detected. In the case of a noisy data (Figure 4) this is a lot harder.
Shot 4
Shot 1 P Shot 2 Shot 3
S
Airwave
Figure 3. Example of a good data set after removing the noisy traces, shot gather.
First P arrival is clear. Also S wave and airwaves are seen. First four shots from the
line 30. AGC has been performed. Trace number on X-axis and two way travel time in
seconds on Y-axis.
8
Airwave
Figure 4. Example of a noisy data set after removing the noisy traces, shot gather.
First four shots from the line 34. AGC has been performed. Trace number on X-axis
and two way travel time in seconds on Y-axis.
The processed and interpreted seismic results were compared and validated to set of
existing knowledge on the site.
The seismic velocities and signal frequency range have been investigated in preceding
seismic VSP works (Enescu et al. 2004). Distribution of velocity data and velocity
anomaly locations in boreholes have been extracted from borehole geophysical
logging, e.g. Julkunen et al. (2004), Lahti & Heikkinen (2005). Petrophysical
properties of Olkiluoto lithology, and the elastic contrasts, were considered during
prior modelling by Mr. Tomas Lehtimäki (Saksa 2006, in preparation) for geometrical
observability, frequency range and attenuation; and by Dr. Christopher Juhlin in
Uppsala for considerations of, e.g., stacking fold and signal to noise ratio.
3 DATA PROCESSING
The aim was to perform reflection seismic processing for two-dimensional seismic
lines, surveyed originally for refraction seismic interpretation. Advantage of this work
is to get three-dimensional reflection seismic results from the existing data at only
processing costs. Processing was based on a processing flow of SKB studies presented
in Juhlin et al. (2001, 2004) and testing of processing to data set of this report for line
S28, performed by Dr. Christopher Juhlin in Uppsala (December 2005).
After processing the seismic sections were interpreted in order to map the orientation,
continuity, location and shape of subhorizontal 3D reflectors as well as to locate
possible subvertical faults.
Before seismic interpretation the survey data must be carefully processed. The
processing steps are presented in Table 2. Line S28 is used here to exemplify the
processing steps (see processing flow in Table 2). Practically no reflectors can be seen
in the raw data from the line S28, shown in Figure 5. After filtering and stacking,
normally some discontinuous traces of reflections can be estimated. Even though
automatic gain control (AGC) is applied to data at the end of processing, it has been
applied to images in order to make the visualization more illustrative. AGC improves
visibility of late-arriving events in which attenuation or wavefront divergence has
caused amplitude decay.
The most important processing steps to bring the reflections visible were refraction
statics and deconvolution. The goal was to find the processing parameters that could be
easily and quickly applied to the whole seismic data set.
Refraction static corrections will remove the differences of travel times in first arrival,
caused by differences in elevation and near surface velocity variation in soil and
weathered layers. Deconvolution will sharpen the signal signature and remove multiple
waves from reflections. During deconvolution, adding noise serves also as frequency
equalization (spectral whitening).
10
Figure 5. Raw shot gather from the line S28 with automatic gain control. First nine
shots on line. Trace number on X-axis and two way travel time in seconds on Y-axis.
Processing was performed using free seismic data processing package Seismic Unix
(SU) (Stockwell & Cohen 2002). Refraction seismic data SEG-2 format was converted
into SEG-Y format before importing into SU (Table 2, item 1), using shareware
software IXseg2segy (Interpex Ltd). Number of samples in SEG-2 format was limited,
and conversion was preferred rather than down sampling. In down sampling some
features of the data could have been lost.
Bad traces to be removed (Table 2, item 2) were selected visually using ReflexW
software (ReflexW) and then removed in SU. Typically these geophones are located
near the shots or the geophones were not properly mounted into the ground.
The next step after removing the noisy traces was setting the geometry (Table 2, item
3). As described in Chapter 2 the geophone spacing and shot spacing are irregular.
Line number S28 consists of 432 seismic traces and for each trace, trace number, shot
number, geophone number, shot position, geophone position, common depth point
(CDP) position and offset must be set. Common depth point is a common reflecting
point at depth on a reflector (see Figure 6). In the case of planar layers, CDP is a
halfway point between a source and a receiver as illustrated in Figure 6. In this work
the CDP locations have been computed assuming that reflectors are planar. Geometry
was read from the ASCII file, which was created with Microsoft Excel, and applied to
data using Seismic Unix. Example of geometry file is shown in
Table 3. Example of a geometry file from the line 28. First 12 traces.
Purpose of this preliminary filtering was to remove the highest and the lowest
frequencies. Selection of parameters was based on a raw spectrum. Example of a
spectrum from the line S28 is shown in Figure 7. The frequency varies from very low
to very high frequencies. Shot gather after the preliminary filtering is shown in Figure
8. Reflectors are not yet visible.
13
Figure 7. Original spectrum from the line S28. Traces 297-363. Selection of
parameters for preliminary filtering was based on this spectrum in order to remove the
highest and lowest frequencies.
14
Figure 8. Shot gather from the line S28 after the preliminary filtering with automatic
gain control. First nine shots. Trace number on X-axis and two way travel time in
seconds on Y-axis.
Refraction static corrections (Table 2, item 6) were calculated using Microsoft Excel
and were applied to data after the preliminary filtering. Values of static corrections
must be integers in milliseconds. The purpose of the refraction static corrections was to
eliminate the effect of the weathering layer and the surface elevation variations (see
Figure 9). Refraction statics calculation was based on a first P-wave direct arrival
times (first breaks, see Figure 3). Refracted first breaks represent the base of a
weathering layer.
SHOT
SURFACE
RECEIVER
WEATHERING LAYER
v = 5000 m/s
Figure 9. Refraction static corrections eliminate the effect of the weathering layer and
the surface elevation variations. Arrows represent the value of correction applied to
trace located between the shot and the receiver.
The static error was removed from each trace by computing the theoretical first breaks
using the replacement velocity of 5000 m/s and trace offset. Correction was obtained
by subtracting theoretical first break from observed first break. After applying
refraction static corrections to seismic data, the base of weathering layer becomes a
15
new datum and elevation differences between the shot and the geophone and between
the geophones are eliminated (Figure 9). Because depth of the real weathering layer
may vary from trace to trace, some differences between the shots might still remain.
We can estimate that these variations are approximately +/- 2 m based on a previous
interpretations (Ihalainen 2003, Lehtimäki 2003), which leads to +/- 0,8 ms error in
two way travel times using the replacement velocity of 5000 m/s. This is acceptable,
because the average amplitude among the traces is many times greater than this error
and because error is smaller than the 1 ms resolution of refraction static corrections.
Comparison between stacked sections with and without refraction static corrections
(Figure 20 a) and Figure 20 b) and Figure 22 and Figure 24) indicates that the
approach is easy and fast and therefore adequate and reasonable to perform. Example
of a shot gather before and after the refraction statics is shown in Figure 10 and
Figure 11.
Figure 10. First three shots with AGC from the line S28 after the preliminary filtering
but without refraction statics. Trace number on X-axis and two way travel time in
seconds on Y-axis.
16
Figure 11. First three shots with AGC from the line S28 after the preliminary filtering
and refraction statics. Trace number on X-axis and two way travel time in seconds on
Y-axis.
Deconvolution (Table 2, item 8) applied Wiener predictive error filtering. Aim was to
recover high frequencies, attenuate multiples, equalize amplitudes, produce a zero-
phase wavelet and generally affect the wave shape. Deconvolution was performed
using the following parameters and is one of the most important processing steps.
Stacked sections were compared with and without deconvolution in Figure 21 a) and
Figure 21 b), respectively. Figures indicate the effect of deconvolution. After the
deconvolution the data was filtered (Table 2, item 9) with similar filter as before
deconvolution, using parameters shown below.
0-100 ms
lower cutoff frequency: 70 Hz
lower plateau frequency: 140 Hz
upper plateau frequency: 300 Hz
upper cutoff frequency: 450 Hz
17
50-200 ms
lower cutoff frequency: 60 Hz
lower plateau frequency: 120 Hz
upper plateau frequency: 300 Hz
upper cutoff frequency: 450 Hz
150-500 ms
lower cutoff frequency: 50 Hz
lower plateau frequency: 100 Hz
upper plateau frequency: 270 Hz
upper cutoff frequency: 400 Hz
Shot gather after the deconvolution and the second filtering is shown in Figure 12.
Now the reflectors after 130 ms and 200 ms are seen. It is hard to see these reflectors in
the shot gathers where the shot point is located between the geophones because of the
airwave.
Figure 12. Shot gather from the line S28 after deconvolution and second filtering with
AGC. Reflectors can be seen under 130 ms and 200 ms marked with red and blue
arrows respectively. Airwave is marked with orange arrow. Trace number on X-axis
and two way travel time in seconds on Y-axis. First 6 shots on line.
The behaviour of records according to shot offset was viewed in common offset
gathers (Figure 13). The airwave (red line) was removed by muting (Table 2, item 10)
at velocity 340 m/s vs offset. The air blast duration was set as a function of offset.
Number of points to taper before hard mute was set to 1000. Common offset gather
18
after the muting is shown in Figure 14. After the muting automatic gain control (AGC)
was applied (Table 2, item 11) to data in order to improve visibility of late-arriving
events in which attenuation or wavefront divergence has caused amplitude decay.
Figure 13. Offset gather from the line S28. Airwave with the velocity of 340 m/s is
marked with red line.
Figure 14. Common offset gather from the line S28 after muting.
The calculated refraction statics were quite good on basis of lining up of P-arrival (see
Figure 12). Some residual errors still remain, which must be corrected for. These
errors can be seen after deconvolution as displacements in first breaks and the reflected
events. The residual statics (Table 2, item 12) are plus-minus errors with respect to the
long wavelength trend of travel time anomalies due to the near-surface effects (Li
1999). Corrections were applied to traces by picking the first break maxima and
computing their differences to the theoretical first breaks, which were computed during
the refraction statics.
19
Original first breaks were picked to zero amplitude (crossover). This work applied
maximum amplitude, so datum was shifted half a wavelength (2-3 ms). After this shift,
datum was 0 m +/- 2 m. This is because before the residual statics datum was 5 m +/- 2
m which corresponds to 2 ms shift in two way travel time. In Figure 15 is shown a
shot gather and in Figure 20 b) a stacked section before the residual statics and in
Figure 16 a shot gather and in Figure 21 a) a stacked section after the residual statics
(see also Figure 23 and Figure 24).
Residual statics makes the reflecting events clearer and straighter. Residual statics
were not applied to the data from the survey lines S33 and S34 because reliable picking
of first break maxima was not possible due to noisy data (see Figure 4).
Figure 15. Shot gather from the line S28 after the muting but before residual statics.
Trace number on X-axis and two way travel time in seconds on Y-axis. First 4 shots on
line.
20
Figure 16. Shot gather from the line S28 after the muting and residual statics. Trace
number on X-axis and two way travel time in seconds on Y-axis. First 4 shots.
Reflections can be seen e.g. at 013 and 0.2 seconds.
Normal moveout (NMO, Table 2, item 13) compensates for the effects of the
separation between seismic sources and receivers and must be performed prior to
stacking. NMO velocity used was 5000 m/s and stretch mute 4. Samples with NMO
stretch exceeding stretch mute were zeroed. Shot gather after the normal moveout is
shown in Figure 17. In Figure 18 is shown a detailed example of the first shot from
the line S28. Originally first breaks and reflecting events are hyperbolic but NMO
straightens them up.
21
Figure 17. Shot gather from the line S28 after the normal moveout. Trace number on
X-axis and two way travel time in seconds on Y-axis. First 4 shots.
22
Figure 18. First shot from the line S28 before and after NMO. Red line demonstrates
how NMO straightens up first breaks.
After NMO data can be stacked (Table 2, item 14). First it must be sorted into
common depth point gathers, which means that data is sorted by its CDP position, and
CDP positions must be changed into CDP numbers. Traces with the same CDP number
are stacked. CDP numbers are integers and calculated based on a position of each
CDP. All traces having their CDP position within a bin length are stacked into the fold.
Length of a bin was 2,5 meters and consequently maximum fold is eight and minimum
fold one. Figure 19 demonstrates principle of stacking. In the line 342 geophone
spacing was smaller so bin is 1 meter. Each stacked sample is divided by the square
root of non-zero values stacked. Final stacked section from the line S28 is shown in
Figure 21 a). Only the first 300 ms of the data are displayed because of a low signal-
to-noise ratio in the lower parts of the data.
23
CDP number
a) CDP position b)
Figure 19. a) Traces before stack b) Same traces after stack. Six traces within 2,5 m
bin (red and green lines in figure a) ) are stacked. Corresponding stacked traces are
marked with red and green arrows in figure b). In reality CDP positions are not
equally spaced and therefore fold varies.
24
a) b)
Figure 20. a) Stacked section without refraction and residual statics from the line S28
down to 300 ms. CDP number on X-axis and two way travel time in seconds on Y-axis.
b) Stacked section without residual statics from the line S28 down to 300 ms. CDP
number on X-axis and two way travel time in seconds on Y-axis.
25
a) b)
Figure 21. a) Final stacked section from the line S28 down to 300 ms using a wiggle
trace display. CDP number on X-axis and two way travel time in seconds on Y-axis. b)
Stacked section without deconvolution from the line S28 down to 300 ms. CDP number
on X-axis and two way travel time in seconds on Y-axis.
26
Figure 22. Stacked section without refraction and residual statics from the line S342
down to 300 ms. CDP number on X-axis and two way travel time in seconds on Y-axis.
27
Figure 23. Stacked section without residual statics from the line S342 down to 300 ms.
CDP number on X-axis and two way travel time in seconds on Y-axis.
28
Figure 24. Final stacked section from the line S342 down to 300 ms using a wiggle
trace display. CDP number on X-axis and two way travel time in seconds on Y-axis.
29
4 INTERPRETATION
After the seismic processing has been performed for the selected set of refraction
seismic survey lines, data was interpreted. First step was to examine two dimensional
stacked sections, which can be found from the APPENDIX 1, and to locate possible
reflectors. After this, two-dimensional stacked sections were imported into the
visualization environment OpendTect (OpendTect 2006) and reflectors were picked in
3D.
Two dimensional seismic data from the survey lines shown in Figure 1, was processed
using the processing parameters in Table 2. The result was a set of stacked seismic
sections which can bee seen in APPENDIX 1 using variable density display. These
two-dimensional sections can be used to locate the reflectors but the actual picking is
done in 3D. Taking a closer look at the reflector below 130 ms in lines S28, S29, S30
and S31, it can be seen continuing from line to line. Very clear reflectors can be seen in
several lines but for example in line S47 only few weak ones are seen. Reflectors are
mostly found from the upper parts of the seismic sections. This is due to the lower
signal-to-noise ratio (see shot gathers, Figure 5) in the later times of the data (greater
depths). This can also be seen from the stacked sections (see APPENDIX 1). Two-
dimensional reflectors are shown in APPENDIX 1.
Some reflectors may not appear to be very clear in first place, and may be hard to
detect from individual lines. Different image colour settings, aspect ratios and scales
were applied. Picking in 3D allows locating reflectors on adjacent and crossing lines,
which helps to recognise continuous reflecting events. In Figure 25 (Lines 48, 50 and
362) the reflectors are marked with red arrows.
In APPENDIX 1 scale is 1:5000 in order to display data more clear. Picking was easier
with stretched X dimension in 3D.
Depth of the reflectors is determined using the approximate velocity of 5000 m/s. The
velocities used in static correction and NMO give a generic idea on the real velocities
in bedrock. However the true velocity has to be confirmed using velocity fitting on
diffractions seen on seismic sections, from external sources (VSP data and 3D
reflection data) and with using a borehole control. Processing has applied a phase
consistent (zero phase) work flow. Correctness of depth axis will rely on recognition of
correct phase of the reflected arrival, e.g. ¼ - ½ wavelength late picks will mean
approx. 5-10 m too deep depth level or the reflection. Selecting the velocity e.g. either
5000 m/s, or 5500 m/s, will lead to approx. 10% error in the depth axis.
30
Figure 25. Example of a weak reflector continuing from line S48 to line S50 to line
S362. It would be hard to locate the reflector from the single lines without the help
from the crossing lines.
Twelve surfaces and 23 single 2D reflectors in total were picked and are shown in
APPENDIX 1. 2D reflectors constructing surfaces are named according to the surface
name and single reflectors with a prefix ‘single’. Surfaces were mainly in focus but
also some single reflectors can be found interesting (see below). Normally the high
amplitude reflectors are not continuous over the whole survey area, but display rather
broken and some times discontinuous character. Also abundant diffracted energy (steep
31
Figure 26. Intersection of lines S45 and S322. Reflectors continue from line to line and
are marked with arrows.
32
Figure 27. Three-dimensional surface picked from the lines S312, S432, S28, S29, S30,
S31 and S43.
The time picked surfaces, recognised from several parallel and crossing lines, were
corrected for offset and offline reflection geometry. The original picks are displayed on
the sections with their apparent dips (see APPENDIX 1). The observations are
geometrically tilted so, that the shortest distance from CDP is measured along the
normal of the reflection plane (reflector is tilted towards the up-dip direction). This
corresponds roughly a 3D migration, but is a simplified procedure (Prissang et al.
2004). Table 4 shows the information on the corrected surfaces. Surfaces are displayed
in Figure 28.
Surface Dip direction/ Dip Two way Depth (m) Picked from the lines
name (degrees from North travel time
and Horizontal) (ms)
ref 1 134,30 N/ 16,18 416 -723,8…-623,3 48, 50, 362, 372
ref 2 275,16 N/ 63,91 178 -199,8…-131,2 50, 362
ref 3 140,32 N/ 18,79 322 -284,5...-251,5 342, 47
ref 5 130,87 N/ 11,51 421 -333,7...-273,9 312, 432, 28, 29, 30, 31, 43
ref 6 109,62 N/ 33,59 374 -499,0...-419,2 28, 29, 29, 30
ref 7 89,22 N/ 10,66 344 -205,8...-162,5 28, 29
ref 8 147,09 N/ 8,29 329 -685,4...-664,1 29, 43
ref 9 442, 44, 282, 45, 322, 312,
115,20 N /17,86 417 -305,2...-192,0 46, 302
ref 10 128,74 N/ 14,52 439 -252,5...-186,7 442, 302, 45, 46, 312
ref 11 138,74 N/ 15,43 423 -298,2...-243,5 432, 322, 45
ref 12 219,34 N/ 18,21 427 -331,7...-252,0 342, 47, 35, 48, 45
ref 13 226,95 N/ 23,26 435 -350,0...-301,4
1. Reflectors ref1 and ref8 are dipping gently to the southeast (134-147/8-16), and
are located at 600-700 m depth level. Their corresponding structure category is
HZ21. Reflectors found 30-50 m above the indications.
2. Almost similarly oriented ref6, intersecting KR29 (485 m), and located 50-100
m above ref1-ref8 or HZ21 (may form a part of it) at depth level 420-500 m
3. Reflectors ref3, ref5, ref9, ref10 and ref11 form a group showing slightly
discontinuous character and varying orientations 128 -140/11-19 in different
locations, and forming upper (ref3, ref10, ref11) and lower surface (ref5, ref9).
Corresponding structures are HZ20A or BFZ098 and HZ20B.
4. Almost similarly oriented ref7 (89/11) at 150-200 m depth levels near KR29, as
projected would be some 40 m above the level of HZ20A, and may indicate
displacement in the level of corresponding structures.
5. Ref12 and ref13, dipping gently to the SW, and met in slightly deviating two
parts in KR22, KR23, KR25 and KR28 at depth levels 320 – 460 m, at 30 m
34
6. Ref2 dipping steeply to the west 275/64, met at 150 m in KR25, and projected
from different distances to intersect at 345 – 517 m borehole lengths in KR4,
KR24, KR28 and KR38. No clear borehole indication. May be difficult to
project at correct location due to steep orientation.
In Figure 28 are shown the picked surfaces and boreholes with bedrock features, e.g.
brittle fracture zones (BFZ, Paulamäki et al. 2006), or hydraulically conductive
features (HZ, Ahokas & Vaittinen 2007) and different lithological units. Reflections
match better to the fractured section locations than the rock type variation.
From the intersections of for example borehole KR07 and two blue surfaces ref9 and
ref10, can be seen that the surfaces match with these features in the geological model
(Paulamäki et al. 2006). These fit well with borehole observations in KR4, KR7, KR24
and KR38 when projected along the plane. Also reflector intersection at borehole
KR25 (ref 12) gets some support from the borehole data (Julkunen et al. 2004).
In Figure 29 is shown a vertical section along the borehole KR07. Red intersection
lines of ref9 and ref10 match well with HZ20A and fairly with HZ20B and their
borehole intersection depths (Ahokas & Vaittinen 2007).
Figure 28. 3D view (from the south-west) of interpreted surfaces, boreholes (KR),
bedrock features and lithological units intersected in boreholes. ONKALO is shown in
yellow.
36
In Figure 30 are shown the features HZ20A (green) and HZ21 (violet) (Ahokas &
Vaittinen 2007 in prep.) and in Figure 31 the brittle fracture zone OL-BFZ098
(turquoise) of geological model (Paulamäki et al. 2006). Surfaces ref10 and ref11
match well with the upper layer (HZ20A) and surface ref1 with the lower layer (HZ21)
in Figure 30 (Ahokas & Vaittinen 2007). Ref 9 in Figure 31 matches with the OL-
BFZ098, and may indicate it could continue towards borehole KR29 (280 – 330 m,
met at 330-331 m). Surface ref5 is located near the borehole KR29 and could be
combined with the surface ref9 and therefore also be part of the OL-BFZ098.
37
Figure 30. 3D view (from the south-west) of interpreted surfaces, boreholes (KR),
HZ20A (green) and HZ21 (violet) and ONKALO (yellow).
38
Figure 31. 3D view (from the south-west) of interpreted surfaces, boreholes (KR), OL-
BFZ098 (turquoise) and ONKALO (yellow).
In Figure 32 the picked surfaces are shown with the results of 3D survey (Juhlin &
Cosma 2006) on stacked sections (same processing phase). The reflection amplitude
maxima are seen on the same arrival times in both sections. The 3D survey suggests
that many of these reflectors seen in 2D lines, are discontinuous in character over
longer distances.
39
Figure 32. Matching events in the stacked 2D (Line S46, see APPENDIX 1) and 3D
sections (for 3D before dip moveout and migration, Juhlin & Cosma 2006). View to the
Southeast.
Evidently the reflecting surfaces correspond to geological structures, e.g., brittle fault
zones met at the borehole intersections. These can be confirmed e.g. from KR4, KR7,
KR25 and KR29 at borehole depths shown in Table 5 above. The discontinuities on the
surfaces suggest slight vertical displacements (faults), though the large scale continuity
of the surfaces on the same depth level can be confirmed.
According to location, ref9 (dark blue in Figures 28 - 32) can be met in KR4, and may
continue on slightly different level to borehole KR29, matching with BFZ098 with
slight vertical offset (330 m). The ref10 (light blue in Figures 28 – 32) is met upper in
the borehole KR7 and would match well HZ20A. The ref1 (red) is fitting well with
HZ21. There is no direct borehole control for the reflector nearby, but evidently the
strongly fractured section in KR4 and KR7 would match this feature. It is not seen
continuously in 2D line sections due to low S/N, but in 3D the event is clear and
continuous.
In Figure 33 is shown line number S46 and boreholes KR4 and KR7 with P-wave
velocity and density logging results. The velocity minima in logging match with the
reflectors marked with red arrows. The reflectors ref3, ref5, ref9, ref10 and ref11 seem
to correlate to this feature, being slightly offset (or faulted) in different parts, and
showing slightly varying orientations 128 -140/11-19 in different locations. Projected
intersection point is in KR4 290 m and in KR7 230 m, slightly above true position, like
at 315-320 m in KR4.
40
Figure 33. Line number S46 and boreholes KR4 (on the left) and KR7 with P-wave
velocity (blue) and density (green), a view from the North. The reflector at 130 – 150
ms time is clearly seen in the image, and will match also the borehole geophysical
velocity minima and fracturing indications (borehole length 220 m in KR7 and 320 m
in KR4). Borehole trace is in yellow; please note the boreholes are off-plane.
41
5 CONCLUSIONS
Aim of this work was to process two-dimensional refraction seismic data collected
from the ONKALO area using reflection seismic processing methods and to locate
gently dipping reflectors from the processed survey lines. This was a demanding task
because of a small fold, approximately only four. Survey lines S28-S35, S43-S48,
S282, S302, S312, S322, S342, S362, S372, S432, S442, S443, S50 measured in 2002
were selected for the seismic processing. After the processing the two dimensional
reflectors were picked in OpendTect visualization environment and combined into
three-dimensional surfaces. Special attention was given for the detection of faults and
discontinuities. The surfaces were transferred to 3D presentation utility and compared
to available geological 3D information (Paulamäki et al. 2006).
Processing was done mainly using free seismic data processing package Seismic Unix
(SU). The most important processing steps were the refraction statics and
deconvolution. Without these steps and careful selections of parameters it was not
possible to see any clear reflectors in stacked sections (Figure 20 a) and Figure 21 b)).
Same processing parameters were applied to the whole seismic data set but refraction
and residual statics needed to be computed separately for each survey line. Residual
statics was the most time-consuming processing step.
Result of this work was a set of two-dimensional seismic sections and three-
dimensional surfaces picked from these sections presented in 3D presentation utility.
Quality of the seismic data was quite good down to the 200 ms (400 m) but after that
signal-to-noise ratio decreases, which can be seen in the seismic sections (APPENDIX
1). Therefore reflectors are mainly found from the upper parts of the data. If the shots
are located near the geophones, airwave covers most of the data down to 200 ms (400
m). Processing was completed successfully, as clear two-dimensional reflectors were
detected from the data. In 3D these reflectors continue from line to line and construct
surfaces which can be compared to borehole observations.
Refraction statics were calculated quite roughly using picked first breaks and
replacement velocity of 5000 m/s in order to get the best possible coverage of the
ONKALO area within the available schedule. In practise the applied procedure led to
adequate accuracy of the results. Another option could have been to use for example a
method called seismic travel time tomography (ray tracing) to determine static
corrections and near-surface velocity variations (Bergman et al. 2004).
Fold was quite small, which has an effect on stacked sections. With a bigger fold
reflectors would have been clearer and have a bigger amplitude, which would have
made the interpretation easier. From the stacked sections (Figure 21 a) and Figure 24)
we can see that reflectors are mainly seen because of their phase coherence.
Dips and locations of three-dimensional surfaces were corrected using the procedure
which corresponds roughly a 3D migration (Prissang et al. 2004). Migrated three-
dimensional surfaces were transported to AutoDesk environment, where they were
displayed with other data. Depth of the reflectors was determined using the
approximate velocity of 5000 m/s. Selecting the velocity e.g. either 5000 m/s, or 5500
m/s, will lead to approx. 10% error in the depth axis.
Because of an airwave, that covers most of the reflectors in the upper parts of the inline
shots, one option would have been to remove all the traces with offsets bigger than
42
some specified limiting value. This could have reduced noise and made the reflectors
clearer, but on the other hand decreased fold. Effective survey would apply longer
offsets to cover steeper reflector surfaces, and more fold in stacking.
The line coverage 100 x 100 m seemed to be sparse to decide on same features on
parallel lines. The crossline control of reflectors allowed detection of several 3D
surfaces, which are limited in the area. More precise mapping of the subsurface
features would require not only higher fold and offset in survey setup, but also denser
line coverage, which demands will practically lead to a need for 3D reflection array.
43
REFERENCES
Juhlin, C., Palm, H. & Bergman, B. 2001. Reflection seismic imaging of the upper
crystalline crust for characterization of potential repository sites: Fine tuning the
seismic source. SKB Technical Report TR-01-31, 49 p.
Juhlin, C., Palm, H. & Bergman, B. 2004. Reflection seismic studies performed in the
Laxemar area during 2004. Oskarshamn site investigation. SKB, Stockholm. Report P-
04-215, 53 p.
Lahti, M. & Heikkinen. E. 2005. Geophysical borehole logging of the boreholes KR23
extension, KR29 and KR29b at Olkiluoto 2004. Posiva Working Report 2005-17, 77 p.
Li, Xinxiang. 1999. Residual statics analysis using prestack equivalent offset
migration. Calgary, Alberta. 141 p. Thesis.
Paulamäki, S., Paananen, M., Gehör, S., Kärki, A., Front, K., Aaltonen, I., Ahokas, T.,
Kemppainen, K., Mattila, J. & Wikström, L., 2006. Geological model of the Olkiluoto
Site. Version 0. Posiva Working report 2006-37, 355 p.
Prissang, R., Hellä, P., Lehtimäki, T., Saksa, P., Nummela, J. & Vuento, A. 2004.
Identification of mineable blocks in dimension stone rock masses. In Hardygóra, M.,
Paszkowska, G. & Sikora, M. (eds): Mine Planning and Equipment Selection 2004.
Taylor & Francis Group, London, Great Britain. p. 69-74. ISBN 04 1535 937 6
44
ReflexW, User’s Manual, Version 3.0, K.J Sandmeier, Zipser Straße 1, D-76227
Karlsruhe, Germany. 345 p.
Stockwell, J. Jr. & Cohen, J. 2002. The New SU User’s Manual. 141 p.
http://www.seismo.unr.edu/
Vaittinen, T., Ahokas, H., Heikkinen, E., Hellä, P., Nummela, J., Saksa, P., Tammisto,
E., Paulamäki, S., Paananen, M., Front, K. & Kärki, A. 2003. Bedrock model of the
Olkiluoto site, version 2003/1. Posiva Working Report 2003-43, 266 p.
45
ref 7
ref 5
ref 6
Single 1
Single 2
ref 9
ref 7
ref 9
ref 6
ref 8
Single 3
ref 5
Single 4
Single 5
ref 6
ref 10
ref 9
ref 5
ref 5
ref 10
ref 9
ref 5
ref 5
ref 11
ref 9
ref 13
ref 13
Single 7
Single 6
ref 3
ref 12
ref 12
Single 8
Single 9
Single 10
Single 11
ref 1
Single 12
Single 13
ref 2
ref 1
ref 1
Single 14
ref 5
ref 8
ref 5 ref 11
Single 15
Single 16
Single 17
Single 18
ref 9
Single 19
ref 10
ref 9
Single 20
ref 10
ref 10
ref 11
ref 9
ref 12
ref 10
ref 10
ref 9
Single 22
Single 21
Figure 44. Reflectors from the line S46.Vertical reflector indicates possible vertical
displacement along a subvertical fault.
89
ref 9
ref 3
ref 12
Single 23
ref 12
ref 1
ref 2
ref 1