Eos, Transactions, American Geophysical Union, Vol. 75, No. 41, October I l.

1994, Pages 473-476

Project Images Crust,

and metamorphic facies changes, all of
which are represented at the well. Seismic
reflections are found both from dipping

Collects Seismic Data in com• positional layering which correlates

with structure, and from horizontal zones,
which are provisionally interpreted as

World's Largest Borehole horizontal, fluid-filled fractures.

Pr*ct
The principal aim of the project was to de-
D. K. Smythe, S. B. Smithson, C. Gillen, C. Humphreys, Y. termine the reflection characteñstics of the
Kristoffesen, N. A. Kamev, V. Z. Garipov, N. I. Pavlenkova, and the upper crystalline crust. Obtaining a whole
Kola-92 Working Group crustal image was a secondary priority.
Since crustal reflection surveys of the
kind now commonplace in the west had
An anomaly in oor understanding of A crucial factor about SG-3, apart from its never been undertaken at the SG3 well, a
per continental crustal structure is caused by depth of 12.2 km, is that the primary lithologi- multinational geophysical team was assem-
conflicting observations from structural ge- cal layering of the Proterozoic supracrustal bled to obtain the necesary surface and
ologists and reflection seismologists. Near- rocks dips at 400—500 (Figure 2). Seismic
downhole seismic data. Reconnaissance site
surface structure in metamorphic terrain is re-
surveys were carried out by the summer of
frequently steeply dipping or vertical, but flections were observed from this layering
1991. The nonhem branch of Environmental,
seismic reflection images display flat-lying re- both on high-resolution shallow surface data
Geological and Geophysical Investigation or-
flectors below depths of a few kilometers. as well as on the previously recorded single
ganized the field work in cooperation with
The seismic image of the topmost 2—3 km is component analogue-record VSPs. There are
usually blank. How can this be, when seis- also flat•lying, lower•frequency reflections be- the Universities of Bergen, Glasgow, and Wp
mic reflections are supposed to represent pri- tween 6 and 9 km depth, cross-cutting the
mary geological structure? terozoic/Archean lithologies (Figure 2). The field work was carried out between
Seismic measurements in and around a Pavlenkova [ 19921 has reviewed the sig- early March and mid-May 1992. Normally, the
deep borehole could solve this dilemma. Sur- nificance of the flat•lying reflectors. The late winter provides the best combination of
face seismic observations and physical sam- depth range at which they occur corre- calm weather conditions, hard ground, and
ples collected from the hole by wireline sponds roughly to the zones of porosity in- enough daylight for seismic surveys. Unfoflu•
logging, particularly by vertical seismic crease, velocity inversion, increase in P wave nately, however, the unusually mild winter of
files (VSPs), could then be compared. anisotropy and pressure, and the presence of 1992 produced very deep snow but barely fro•
There are only two sites where hypothe- circulating fluids. Furthermore, the essen- zen ground.
ses on crustal reflection sources can be tially vertical variation in physical properties First, a 38-km long crustal seismic reflec-
against in situ geological and fluid sam- correlates well with discrete changes in meta- tion profile was recorded through the well in
ples. One of these sites, the Kola morphic facies. a direction at right angles to the regional
super- The SG3 well provided an ideal opportu- strike. The surface and VSP data were re
deep well in the Kola Peninsula, northwest nity to test hypotheses on several alternatives corded in three•component mode—ground
Rusia, has been cored with more than 80% to primary lithological layering as the source motion in x, y, and z directions—lo give the
recovery. Investigation has yielded exciting of reflections. The alternative sources of crus- full wave field, permitting separation of Pand
evidence of current and past fluid flow, as tal reflections include shear zones, fluids, Swaves. A surface seismic survey was then
well as physical property and logging data.
However, until now it has lacked the deep
crustal commondepth-point (CDP) seismic
section required for correlation efforts. To
remedy this gap in our knowledge, a major
multinational experiment, incorporating
truck•mounted vibrators, was performed dur•
ing the winter of 1992 (Figure l).

D.K. Smythe, Ikpartment of Geology and

Apptied Geology, University of Glasgow. Glas•
gow 012 8QQ, United Kingdom; S.B. Smith-
son and C. Humphreys, Department of
Geology and Geophysics, University of Wyc»
ming, PO Box 30(h, Laramie,
82071 ; c,Gil-
len, Centre for Continuing Education,
University of Edinbugh, I I Buccleuch Place,
Edinburgh EH8 9LW, United Kingdom; Y.Kris-
toffersen, Seismologi€al Observatory, Univer-
sity of Bergen, Allegaten 41 N-5W7 Bergen,
Norway; N.A. Karev, Rudgeofizika, Fajan-
sovaja 20, St. Petersburg 193019, Russia;
V.Z.Garipovt Ru.s.skomgeologia, B. Gruzin•
skaya 4/6, Moscow 123242. Rusia; N.l. Fig. l. Russian Caterpillar tractors are towing Bergen University's truck•mounted vibrators over
Pavlenkova, Institute of Physics of the Earth, the last hitl on the CDP line. The SG3 well towerlies I km to the north. This image is a Still frame
B. Gruzinskaya 10, Moscow, Russia from a video of the pmject.

Copyright 1994 by the

American Geophysical Union
 Geological cross-section
North from well (km)
1
4
3 4
7 While the CDP cat)le spread progres.scd
8
10
11
Fig. 2. Geological cross-sectton through ,SG3,
showing a oanety ofsouthdipping lithologies.
Bold solid lines indicate seismic reflectors in-
terpreled from previous Russian experiments.
recorded in
formats.
three•componcnt and thrcc-dimensional seismic data, inany timing and internal noise
The main line was Shot from south to
nonh. along a slraighl track- Away froln
the well the maximum lateral deviation of
shots or receivers frolil a Inean straighl
line was :ess than
m. The line coul(i not be ex-
tended more than 3—4 krn north of the well
because ot the
of Ihe Illiliaarized
border zone. South of this zonc iic a
of east•west trending Qbstacles such as
major 50ads, railroad tracks, and
power lines. all of which may cause
no•se and degrade (lata-
Early plans to use dynamiteas a sourcc
were ruled out due ta logisticai problems and
c.ost, so four truck-mounted vibrators were
used instead (Figure I Although truc.k-
gnounted vibrators are hardly ideal for the ler-
rain, they are power(ul. 'Chey are also equpped limited three-dimensional survey was
with control electronics compat• ible with the
recording equipment being u.sed to collect data. carried out around Ihe well instead. ñis
The trucks were each towed on the main line by three-dimen• sionat. threc-componeot
tractors. and three ways of mounting Ihe trucks coverage of a CDP area covers about 6 km
on specia'ly con-slructed skis were tried out.
Duc to mechani-cal breakdowns, thc normal
source tumed out to be three out of the four
vibrators.
Once the convoy of source vehicles started
on Ihe line (Figure l) there was no way lo
tum the vibrators around; this constrained
the possible shot-receiver geometries.
Three 96-channel recording syste[ns were
set up to record the vibrations. Eac.h sysle•n
had its own spread in a 3 x 30 channel mode
(Figure 3a) at channel spacingt giving a
composite spread 4.5.km long. A recording
cabin—or doghouse—with two systems sat
between two spreads. Anothcr doghouse
Irig• gered sourc.e•firing and the recording
cycle of all Ihree instruments. The vibrators,
in an offend configuration south of the
cable, moved up 1.5 km as far as thesouth
end of the Inost southerly spread vibrating
every 50 'n. This spread was then lifted and
laid again in the nonh. Although each spread
was made up of 1'20-channel roll-along
Arctic cable sec-tions in pristine condition, it
was impossible to use it in a convenlional
roll•along mode, in which 12-'24 channels
arc moved at a time. when combined with
three.component re-cording and thrce
separate recording sys-tems.
to• Ward the well frorn aboul 30
south (Fig•
ure 3b), a .separate 5-km long, Oinslrument
Rcnek remote instrument array was rolled to-
ward the well at half the rate of the cable
spread- The two spreads converged and over-
lapped near the well. 'Ihe Refleks operated
within time windows during Ihe day.
"1'0 interface with the Refteks. the
vibrators and geosource MDS- 10 recording
systems had to be starled in absolute time.
Thus total subsurface coverage was 60- to
70-fold. Be• cause it was the first time that
such instru-rnents had been uscd lo record
such vast quantllies of controlled source
problems had to be resolved. The data
collec.ted ex-tend the range of CDP oífsels to
between 10 and 20 km (Figure 3b) and Will
evenloally be merged with the shorter-ofíset
CDP cable spread data.
Extensive wave test; were carried to
compare horizontal and vertical vibrations
collec.ted by acous\lC dctcctors with single
•,hrcc-componcnt cases. Even when Ihe
three strings had been properly planted and
aligned, snow meiling around the phones
soon deteriorated the geometry. En the end,
only Ihe vertical component string buried in
shallow holcs in thc snow was used. along
with two horizontal components taken from
one deeply buricd three•component case.
Since the proposed east-west cross-
iine Ihrough Ihe well was impossible to
observe due to thc weathcr conditions, a
arcnjnd Ihe well, with of(sets from 50 m to
5 km. Thus within each
square arca, three to four traces can
be summed to improve the signal (Figure 3c).
The result is a unique Ihree-dimensional, three-
component crustal reflection data set, complete
Wilh well control to 12 km. 'Ihe ap-enure and
densily of the data are far too small for
orthodox Ihree-dimensional proc-essing, but
the data shouid permit a test of
the hypothesis that there are horizontal
reflec-tors al the well (Figure 2).
It was impoxsible to acquire the VSPs origi-
nally planned. The 12•km long logging cable
wa.s damaged beyond repair, so in.stead sev-
eral 6-km deep, three-component VSPs were run
at different offsets using a new Russian
three•level, threecomponent tool that digit-izes
down•hole (2 m/s sample interval) onto a three-
conductorcable. The VSP calibrates surface
seismic data against the geologic data from the
borehole. The Russian tool was used in a
single-levet mode at a level interval of 20 m,
with a sweep of 12—120 Hz. Up to three
vibrators per source point were u.sed.
Positioning was based on a high-
quality topographic map enlarged to
l: scale. Three-dimensional global pc»
sitioning satellite fixes were made at key loca-
tions. Absolute positioning is better than 20 m
and relative positioning is 5 m or better, which is
satisfactory for processing purposes.
Quality
The south end of the line, Shot first.
shows excellent data (Figure 4a). Lower
crustal re-ilectivity and a Moho—the
boundary surface that separales the Earlh's
crust from the man-lle—at 12 s ('10 km depth)
could be seen on some raw Shot gathers.
Quality of the horizon-tal components is
much lower than the verti-cal component, but
presumed shear wave arrivals can bo secn.
Because the Reflek (lata had never been
ased in this mode before, they required
heavy low-leve( processing to apply time
cor-rections. They also suffered from
intemal problems and 18 Hz and 50 Hz
pickup from pu:nps at the well; however,
good data are being recovered. The quality
ot the VSP data (Figure 4b) is excellent.
The con)plete wave field (in a three-com-
ponent mode) was observed, and all raw data
were preserved in unsummed, uncorre• ünted
form. After demultiplexing, the surface
seismic (CDP) data amounts to some 40 Gbar
and the VSP data to about 2.5 Gbar- A great
deal of data from the second-hand field tapes
had been left in unheated, uninsulated dog•
houses overnight, during whic.h the air tem-
perature sometimes fell to -300C. One
operator•s log sheet even records "Ice on
tape!" However, the data on the corrupt tapes
were succe,xsfulty recovered by reading
thelD back on a field recording system linked
to a PC More than 99% of the
Shot re-
cords of the survey were saved.
Future Investigation
Because the data are presemed in both
uncorrelated and unsummed form, filteñng
methods now being developed can later be

a due to lithological contrasts.
South Composite CDP cable spread Nonh
Vibrators w continental crust; for ex-ample, What is the
30 x 50 m 30 x 50 m 30 x 50 m
b that has been measured in the hole.
Cable spread Refiek spread Well
Vibs 4.5 km 5.0 km
-30 km
Fig. 3. a) Layout oí composite CDP spread of
90 three-component stations linked to sledge-
mounted doghouses W (Wyoming University)
and G (Glasgow University). Vibrators moved
up in an offend, shot•receioer configuration.
b) Reftek remote recorder spread oí 50 three
component stations at 100 m spacing is about
haláoay betunen the cable CDP spread and
the well, and is rolled to 'he north at hall the
rafe. c) threedimensional, three-component
subsurface cove«zge amund the Kola super-
deep well (star). Shaded area around the u,ell
shows depth points for horizontal reflectors.
Crosses are schematic locations of sources.
solid line is locatton o/ the receiver spread. Ar•
mu.) shows continuation oí main line. Gñd is 1
km 2
*ipplied to remove unwanted signals, such display can be dynamical ly switchod
as Ine direct arrivats and the 19Hz hurn from through the trace gather so that adiacent
machinery at the well site. Nonlinear effects vec-tor traces can be co:npared.
of the vibroseis.F,arth interaction, Despite the political, weather. and
fund-ing problems, the
producing harmonics of Ihe sweep. can also
phase oi
be ana-lyzed, and perhaps removed. I he Q.x- through difficult economic times. The west-
Novel methods are also being applied ü)
Deriment was succexsful. The seismic data
three-component data, treating (x,yz_) trace oblained wilE addrc\s the conllict between the the Earth Sciences Division of the National
fripleLs as vector traces rather than within steep lithological dips seen at outcrop and
convcntional vertical (z), iniine (x) and the flat•lying reflectors commonly ob-served Ining (Jniversity) and the Norwegian Re-
(y) gathers, the industry norm wherever deep sei5111ic reflection pro-files
[ T01ham and McCormack, 19911. A new com- are shot. The primary lithological layering in
puter program simultaneously anitnates a bolh the Proterozoic volcanics and the
ground motion and a trace-c,omponent dis• Archaean is steeply dipping—but not, as in
the KTB deep borehole in Ger-many, too
play. whilc the data are rotated about any stcep and complex to image seis-mically. If
. we observe flal-lying, or {ow-dipping
reflectors, as is usually the case with deep
axis. The animation supplies a convincing reflection profiling, then we can re-
Ihree-dimensional illusiont and color coding
and labeling permit identification of Ihe túne
dimension. At the same time, the animated
fute the hypothesis that such layering is
This unique data set should aid our
under-standing of the relative role of
lithological lay-ering, fractures, and fluids in
giving rise to reflectors within the
structural control on fluid flow at mid-upper
crustal depths, and the na• ture of the brittle-
ductile transilion? We should al.so be able
to observe P wave and S wave anisotropy,
and correlate them with the in situ stress
Going down beyond the reach of the drill,
the experiment Will help to define the nature
of the Precambrian Moho, and whether
Archean lower crustal structure is fundamen-
tally different from younger crust, as is now
being suggestcd after the reevaluation of
seis-mological data [Durrheim and Mooney,
1991; Mooney and Meissner, 1992]. It also
provides a crucial section of a crustal
reflection tran-sect across the Ballic Shield
from the Gulf of Bothnia, whcre the BABEL
data already exist [BABEL Working Group,
1993) to the Barents Sea. Conil)ieting the
crustal transect should be rclativcly easy,
since the remamdcr of it could be done
entirely on roads. A short ex-tension
northward within Norway, combined with
on one side
and recording on the other side of an
inacces-sible area—of the no-go zone just
north of the well, would also link into deep
marine reflec-{ion data in the BarenLs Sea.
3 to 15 km, dne well Will be maintained as
a natural
for internationai coopera-
ave experimenls- Using the sevcral supcr-aeep
holes. l.»articularly SCr3. within Che (or-Iner
Soviet I Jnion for a variety of new clown-hole
c.xporiments i5 a cost-cffcctivc way to
su pport the new intcmational continental deep
dri Iling program currently in develop[nent.
of a VHS video showing scenes
of the operations at the SG-3 wellhea(l can
be obtaincd fronl D. K. .S:nythe (e-mail
dks@uk.ac.glasgow.geology).
Acknowledgments
We thank Goskotngeologia (now Russk-
()lngeologia.) for it5 su l)port of this projec.t
ern partner.s were financed by grants from
Science Foundation. EAR-91186(1), (Wyo.
search Council for Scienc.e and the
Hurnanities (Bergen University). Financial as-
sistance witb planning and transport was
given to Glasgow University by the C.arnegie
Trust for the University of Scotland, thc Royal
Society, and the University of Wyoming. We
were also aided by financialt hardware, and/or
software (lonations from Advance Geophysical
Corpora- tion. BP Exploration.
 aO o b De*h {m) Conoco, Enterprise Oil Norge Ltd, Gecc»
stEX, Prakla, Halliburton Geophysical Services,
0.0 and Simon•Horizon Ltd.
1 1

Kola VSP•I: horlmnW

trmowr»

0.5 Reference
BABEL Working Group, Integrated seisrnic stu&
is of the Baltic shield IBing data in the
CAJIf of ü)thnia region, Geophys. J. Int..
1.0 '12, 305, 1993.
Durrheim, R. J. , and W. D. Mc»ney, Archean
and Prc*erozoic crustal evolution: Evidence
3 3 from crustal
Geology, 19, ü,
1.5 1991.
W. D.. and R. Meissner, Multi-ge
netic origin of reflectivity: A review of sei9
4 4 mic reflection pr(filing of the continental
lower crust and %ho, in Continental
2.0 Lower Crust, edited by D. M. Fountain, R.
Arculus, and R. W. Kay. Elsevier, Anster-
dam. 45, 1992.
5 5

Pavlenkova. N. I. , Ihe Kola supedeep drill-

2 3 6 hole and the nature

4 5

Fig. 4. a) Bmte stack ofvertical component data from 'he south endofthe CDP line Inyering at
of sekmic

018tatEe (km) 2.5

tites that have been overthrust to the north. A Terra Nova, 4. 117. 1992.
single oelocity Tatham, R. H. , and M.
2—3 s tu,o-way travel time is from Pechenga oolcanics underlying Archean gneisses and mign•uy D. McConnack, Multi-
function and component seSmolog in
explora-
uve applied, but no static conections or bad trace editing: raw data qualitysimplifiedsthergeomforetrygood. tion, Investigations in Geophysics Series,
b) Example ofa VSP section at short offset without wave field separation. Arrow indicates strong vol. 6, Society of Exploration
downgoing S waves generated by mode conversion at obout 2800 m depth. Tulsa, 248 PP.. 1991.
Kola-92 WorHng Group
C. Gillen, D. K. Smythe, D. C. Abensour, D.
R. Watts, H. Geiger, S. B. Smithson, C. Hum-
phreys, N. Boyd, M. A. Olson, L. Libeny, R.
Blenckner, A. Tanner, R. Carbonell, Y. Kristof.
fersen, C. Aranda, P. Digranes, D. Snanemo,
A. Nilsen, B. Sturt, R. Busby, L. L. Tamovet-
sky, S. Zubkovsky, M. Zarhin, V. Sarakhanov,
Y. Rospolozhensky, V. Stupak, N. l. Karaev, G.
Mikhailov, A. Ronin, N. l. Pavlenkova, D. Ilyin-
sky, S. Malias, l. Morozov, F. Mitrofanov, M.
Brodskaya, M. Strelkova, A. Grishko, D.
Guberman, and M. Lisinsky