139
Tecfonoph_vslcs, 195 (1991) 139-149 Elsevier Science Publishers
B.V
. Amsterdam
european
Deep_ structure of Fennoscandia from fundamental and higher mode dispersion of Rayleigh waves Gildo Calcagnile Dlpurtrmento dt Geology
e Geojwca. Unruersrtci dr Barr, Ban, It&, and Osservatorro dl Geofurca e F~isrccr Cosmrca, Facoltd dl Screnre, Unroersctri dr Ban, Ban, IIO(~ (Rewed
version accepted
geotraverse
Apnl 20.1990)
ABSTRACT Calcagnile, G., 1991. Deep structure of Fennoscandia from fundamental and lugher mode dispersion of Rayleigh waves. In: R. Freeman. M. Huch and St. Mueller (Editors), The European Geotraverse. Part 7. Tectonophysxs, 195. 139-149. Surface wave dispersion for the Baltic Shield area has been Investigated through analysis of phase velocltles up to periods of about 250 s for the fundamental Rayleigh mode and about 80 s for the first two higher modes. Tlus has enabled investigation of the lithosphere-asthenosphere system and deeper regons down to about 400 km. The shear wave velocitydepth relationship obtained from the inversion of dispersion data for the paths COP-KEV/COP-KRK can be considered as representative of an average model for the Baltic Shield. Moreover the mvewon of dispersion data supplemented t,y prewous results from the Scandmavian area suggest the presence of signifxant lateral heterogeneity down to depths of abcut 350 km moving from the older Precambrian shield (COP-KEV/COP-KRK) to the penpheral area (NUR-COP). Models wth small. If any. differences in average shear velocity are consistent with the data m the depth range 350-450 km for the regions studled.
Introduction
preferred even if by this procedure the accuracy obtainable in phase velocity determinations is less than the accuracy proper of dispersion relationship obtained by stacking. The latter process actually leads to a less ambiguous interpretation of
Every major plate, except for the Nazca and Pacific plates, which are mostly oceanic, includes a Precambrian shield. The analysis of available dispersion data indicate that these shield areas have a deep, relatively rigid, high-velocity root extending to depths of at least 200 km. By using
complicated responses and to a better phase velocity determination for modes that are not dominant in amplitude
stacking techniques, Nolet (1976, 1977) has obtained dispersion relationships for the first seven
In previous Clymer, 1973; Calcagnile and
Rayleigh modes in western Europe, inferring structural information down to about 800 km. In a re-analysis of these data, Cara et al. (1980) divided the whole area in a southern
and a northern
However, since western Europe heterogeneous area (e.g. Panza
part.
is a tectonically et al., 1980a,b),
0 1991 - Elsewer
Science Publishers
and Panza,
1976; Nolet,
upper
investigations (Noponen, 1966; Neunhbfer and Guth, 1975: Panza, 1978; Stuart 1978) the
crust
and
mantle
structure
paths were
of the Fennoscandian studied by analysing
under
several
area (Baltic fundamental
Shield) mode
surface wave dispersion with the two-station method. The highest measured period in those studies was around 150 s, allowing the derivation of elastic properties to depths of about 200 km. In view of the importance of providing useful constraints on hypotheses of geodynamic processes such as glacial rebound and plate dynamics, we
with heterogeneities extending down to depths of at least several hundred kilometres (e.g. Calcagnile and Scarpa, 1985). the previous results should be considered as gross regional averages. Thus the use of the standard two-station method, which allows the sampling of more limited areas, may be 0040-1951/91/$03.50
(e.g. Nolet
1977).
B.V
have tried velocity
to extend
to higher
measurements
modes of Rayleigh
mental
of fundamental
the phase and higher
waves. The results of the inver-
sion of phase velocity longer periods
penods
measurements
than obtained
carried
out at
up to now for funda-
and higher modes should allow the acquisi-
tion of more information
on the deeper
parts
of
the mantle. Data In this paper
we report
the results
obtained
with the two-station method (Bnme and Dorman, 1963) applied to a set of events recorded at WWSSN
Fennoscandian
stations
coordinates and earthquake ters are given in Table 1.
(Fig. 1). Station
hypucentral
parame-
For the measurement of higher Rayleigh modes (Panza and Scalera, 1978) the records were digitized at 2 s intervals and processed with a multiple narrow-band
filter system
(e.g. Levshin
et al.,
1972) to obtain time-energy diagrams. However, knowledge of the group velocity tained
in such a manner
fication
is not adequate
of higher modes.
Possible
100
200
Fig. 1. Long-period seismograph network and station lines used to determine phase velocity profiles of Rayleipfi waves.
ob-
for identi-
erroneous
iden-
tification of the modes can be solved by comparing phases that have been determined using frequency filtering and time windowing against
TABLE
I i
theoretical
diagrams
of the phase dispersion
of the
first modes. The derived phase velocity relationships are shown in Fig. 2. Dispersion curves for the same
1
Coordinates
of earthquakes
and recording stations
Location
h (km)
Origin
Time
Olh 26m 36.8 s
Earthquake hypocentral data Kuriles
43.2 N,
146.2 E
76
23 June, 1964
Honshu
40.7 N,
142.9 E
41
29 March, 1965
10h 47m 38.4 s
Hokkaido
44.2 N,
145.5 E
159
25 Oct., 1965
22h 34m 22.4 s
Taiwan
24.9 N,
122.6 E
102
1 July, 1966
05h 5Om 38
Shikoku
33.3 N,
132.3 E
48
5 Aug., 1%8
16h 17m 05.5 s
Hokkaido
44.9 N,
143.2 E
238
19 Jan., 1969
07h 02m 07.9 s
Burma
23.1 N,
94.7 E
124
17 Oct., 1969
Olh 25m 11.5 s
Seismic sratrons Copenhagen
(COP)
55 41’OO”N,
12 26’OO”E
Kevo
(KEV)
69 45’19”N,
27 00’24”E
Kirkenes
(KRK)
69 43’27”N,
30 03’45”E
Kongsberg
(KON)
59 38’57”N.
09 35’54”E
Nurmijarvi
(NUR)
60 30’32”N.
24 39’05”E
Umea
(UMW
63 48’54”N,
20 14’12”E
s
DEEP
STRUDURE
OF FENNOSCANDIA
FROM
DISPERSION
OF RAYLEIGH
WAVES
$ ‘5 8
s
DEEP
STRUCTURE
OF FENNOSCANDIA
FROM
DISPERSION
OF RAYLEIGH
k
WAVES
.I
0..
.n
143
path
obtained
agreement,
from
while
different
events
dispersion
paths are in general
curves
systematically
are in good for different
different.
ever, we must note that. taking
KRK, which both cross longitudinally almost the same area of the shield and are almost equal in
How-
into account
length.
their
In the inversion
dispersion
values
step the average
will therefore
be used.
of their The re-
error bounds the curves are not completely resolved from one another: single “dispersion areas”
sults for the path COP-UME are accounted for if we observe that this path is located in the south-
partly
ern part between
overlap,
particularly
for
the
first
higher
For the fundamental ues in the period KON-KEV the shield.
the previous
As far as higher
mode.
path That
mode,
relatively
range 60-130 bordering might
s are found the western
be explained
no comparable
low val-
modes
data
paths.
values
are available
are concerned, for the area.
for the
Such modes might be compared
side of
by Nolet (1976, 1977) and Cara et al. (1980) with different
by a mixed
into
path of 50-80s shield structure and 50-20% younger Caledonian structure, as suggested by
techniques.
account
that
However, the modes
with those found it should
be taken
obtained
by these
authors are averaged over an area including the Baltic Shield and north-central Europe and this area is very heterogeneous (Panza et al., 1980a,b;
Clymer (1973) for a similar path. The lowest values are found for paths COP-NUR and KONNUR, which is to be expected since these paths lie along the southern edge of the shield. These values are, comparable to those obtained for north-central USA by Biswas and Knopoff (1974). Higher val-
Calcagnile and Scarpa, 1985). Higher mode dispersion relationships seem to be more different for the second than for the first higher mode; the second higher mode values for
ues are found
NUR-COP
for the paths COP-KEV
and COP-
at long periods
are lower than
TABLE 2 Phase velocities (H.M. = higher mode) Period (s)
Error (km/s)
Phase velocity (km/s) Fundamental
I H.M.
II H.M.
CF
(I
CII
(a) COP- -NUR 277.7
4.90
0.12
250.0
4.72
0.12
166.7
4.37
0.06
100.0
4.20
83.3
4.15
62.5 50.0
4.08
41.6
4.04
0.06 5.60
0.06
5.25
5.91
5.03
5.60
0.06
0.12 0.10
0.12
0.10
0.12
0.06
31.3
3.90
4.72
5.13
0.07
0.10
0.10
25.0
3.79
4.63
5.09
0.11
0.12
0.12
0.12
(6) COP-KEV
and COP-KRK
250.0
4.74
0.12
208.3
4.62
0.10
125.0
4.34
0.06
100.0
4.28
71.4
4.23
5.41
6.37
0.06
0.12
50.0
4.10
5.03
5.56
0.06
0.10
0.10
41.6
4.05
4.90
5.35
0.06
0.10
0.10
31.3
3.90
4.73
5.07
0.10
0.10
0.10
25.0
3.79
4.68
4.93
0.12
0.12
0.12
0.06
those
DEEP
STRCICTURF
OF FENNOSCANDIA
for COP-KEV/COP-KRK, previously,
taking
An example ments
of the quality
3 for relevant are
plotted
COP-KRK accuracy
the
and COP-NUR; are
obtained
from
of the data is given Single
paths
the estimates the
spread
for each single period,
the
error
given in Table
of the
with the rms error
of
we assumed measurement,
which was larger than the scatter among together
of the
small number
of the single
Average
phase velocity
the single
for the curve
2.
values of the whole data
set (fundamental
and higher
COP-NUR
COP-KEV/COP-KRK
and
modes)
for the paths which
are used in the inversion step are given in Table 2 together with the associated errors estimated on the basis of the spread
COP-KEV/
measurements
values.
measure-
145
WAVES
if as we said
resolved.
phase data. Due to the sometimes estimated
OF RAYLEIGH
their error bounds
profiles.
for
DISPERSION
even
into account
the curves are not completely in Fig.
FROM
that,
in the data. We stress again
also in this case,
considered
together
phase
with
give two “dispersion
their
areas”
Obviously
it cannot
differences
in the final models
by
different
However. “absolute”
velocity
ing cross sections
in Table
2.
overlap.
ruled
out
that
can be determined
assumed
as we have already
if
partly
be generally
arbitrarily inversion
error
that
values.
cross
sections.
said, ours is not an
but an inversion
are constrained
whose start-
by ea-lier results
Inversion
(mainly Calcagnile 1982; Guggisberg.
The inversion was carried out using the hedgehog procedure (e.g. Biswas and Knopoff, 1974).
Berthelsen, 1987). We therefore believe that the results of the inversion are of some significance. The S- and P-wave velocity and density m the
We are aware of the heterogeneity present uppermost 200 km of the investigated
in the area
(Calcagnile, 1982; Guggisberg, 1986). At this stage, however, we wish to focus our attention on the deeper parts of the upper mantle; hence we inverted phase velocity data for two selected paths. These correspond to two regions, each of which, according to previous studies, may be considered as being relatively homogeneous (Calcagnile, 1982). Actually, looking at the lithosphereasthenosphere system map (or at the lid map) given by Calcagnile (1982) one can see that COPKEV/COP-KRK is a mixed path of more than 80% shield structure with a lithosphere thicker than 150-170 km (and low, if any, contrast between the lid and the low-velocity layer). In contrast, COP-NUR with lithosphere
is more than 90% in a region 120-130 km thick and there is a
clear low-velocity layer lid contrast. This is also supported by Guggisberg and Berthelsen’s (1987) results. All the other paths are more complex. Thus we believe that the path COP-KEV/COP-KRK can be considered at this stage as representative of the average deep model of the Baltic Shield. A detailed analysis of the deep lateral heterogeneity in the area will be the subject of a forthcoming paper.
different
mantle
and Panza, 1978; Calcagnile, 1986; and Guggisberg and
layers
have been
fixed at com-
monly used values and by taking into account the FENNOLORA results (Guggisberg, 1986). Small differences from the quoted values respect have a significant influence
do not in this on the results
of the inversion (e.g. Knopoff and Panza, 1977). The cross sections used in the inversion are given in Table 3. The average crustal thickness in the area is Moho candia, 45 km
about 40 km: the greatest depths to the are found under northeastern Fennoswhere a fairly uniform thickness of about with a maximum in the Bothnian area of
about 50 km is observed. In contr.lst. crustal thicknesses of around 30 km have been found in the coastal area of southern Fennoscandia (Bungum
et al., 1980). Accordingly
we have fixed
at about 42 km the crustal thickness for the path COP-KEV/COP-KRK while for COP-NUR it has been fixed at 35 km (small fluctuations in crustal parameters do not significantly affect the inversion results for the periods that we are dealing with). The fine structure of the subcrustal layers down to about 90-100 km. based on Lund (1979). Cassel and Fuchs (1979) and Guggisberg (1986), was taken into account both for the path COPKEV/COP-KRK, which is almost coincident with the Fennolora profile, and for COP-NUR, which occurs on the edge of the shield.
TABLE
3a
Hedgehog
model for upper
mantle:
COP-NUR Layer thickness
Depth (km)
(km)
Y (km/s)
P Wcrn’b
r’r (km/s)
_-_
.._-_
__
0
crust plus fine subcrustal
layers
90 hd
P4
P2
8.5
3.35
P5
P3
8.4
3.50
P6
Pl
8.8
3.65
490-P4-P5-P6
Pl
9.8
4.00
90
5.2
10.0
4.10
100
6.1
11.15
4.40
300
6.35
11.7
4.60
CQ
6.50
12.0
4.70
90 + P4 channel 90 + P4 + P5 subchannel 90 + P4 + P5 + P6 “spmel”
layer
580 670 770 1070
Confidence Parameter
limits 0 = 0.065 km/s;
single point rejection
if I6C 1 > c (Table 2)
Range
Pl (km/s)
4.65
(0.10)
5.25
P2 (km/s)
4.35
(0.15)
4.80
P3 (km/s)
4.30
(0.10)
4.80
P4 (km)
15
(30)
75
P5 (km)
35
(70)
245
P6 (km)
15
(60)
195
P7 (km/s)
4.70
If in the inversion the difference between computed and experimental phase velocity is larger than _tt (Table 2) at each individual period, the model is rejected. For models that pass this test we check whether the standard deviation (rms) is less than a fixed value (in our case 0.065 km/s). The inversion result is summarized in a simplified way in Fig. 4. The models obtained for the path COP-NUR agree in the uppermost 200 km with those already published (Calcagnile, 1982). Moreover, the bottom of the low-velocity channel is found around 350 km; beneath, the subchannel, which has a shear velocity V, of 4.80 + 0.15 km/s, reaches an average depth of about 500 km and overlies a
(0.20)
5.10
layer with V, = 5.0 + 0.1 km/s corresponding to the “spinel” layer. For the region corresponding to COP-REV and COP-KRK, the layer that we have named “channel” [in the cross section (Table 3) it is formed by two layers separated by the lid] represents an average structure that is similar to that already given by Calcagnile (1982). The average shear velocity in the “channel” is 4.6 & 0.1 km/s down to about 210 km where it decreases slightly for the next 20-80 km. This supports Sacks et al’s. (1979) finding on the depth of the lithosphereasthenosphere in the Gulf of Bothnia area based on body wave studies. Below this layer the shear velocity increases to 4.7-4.8 km/s down to about
DEtP
STRUCTUKE
TABLE
OF FENNOSCANDIA
FROM
DISPERSION
OF RAYLEIGH
WAVES
(km)
V, (km/s)
147
3b
Hedgehog
model for upper
mantle:
COP-KEV/COP-KRK Layer thickness
Depth (km)
P (g/cm3)
VP (km/s)
0 crust
fme aubcrustal
plus
105 channel
layers
1
P5
P2
8.2
3.40
P5
P3
87
3 50
P6
P4
8.5
3.50
P7
Pl
8.8
3 65
475-2P5-P&P7
5.0
9.x
4.00
90
5.2
10.0
4 10
100
6.1
11 15
4.40
300
6.35
11.7
4.60
M
6.50
12.0
4.70
105 + P5 hd 105 + 2P5 channel
2
105 + 2P5 + P6 subchannel 105 + 2P5 + P6 + P7 ” spIneI” layer 580 670 770 1070
Confidence
hmlts u = 0.065 km/s;
smgle point rejection
If I6C 1 > c (Table
II). Different
values (4.9 and 5.1 km/a)
for the “spmel”
layer were also tested Parameter Pl (km/s)
4.65
(0.10)
5 25
P2 (km/s)
4.3
(0.10)
4.80
P3 (km/s)
4.5
(0.15)
4 80
P4 (km/a)
44
(0 10)
4 70
P5 (km)
30
(10)
P6 (km)
55
(20)
70 95
P7 (km)
15
(60)
195
450 km. In the deeper regions the average shear velocity is similar to that for COP-NUR, and models with small, if any, differences are con-
also made it possible for the first time to measure higher mode phase velocity at periods useful for investigating deep mantle structure.
sistent
Analysis of the fundamental and the first two higher Rayleigh modes has allowed information to be gathered on the upper mantle structure down to depths of about 400 km. A possible average mode1 has been obtained for the deep structure of the Baltic Shield. Moreover, the inversion of phase velocity data together with earlier results indicates
with both data sets.
Conclusions The analysis of Rayleigh wave dispersion has allowed phase velocities in the Fennoscandian area to be measured up to about 250 s for the fundamental mode for several new paths and to be extended to much longer periods than was previously possible.
Moreover,
this type of analysis
has
that Fennoscandia is characterized by a significant lateral heterogeneity not only in the uppermost 200 km as already found in previous studies, but
148
StiEAR VELOCITY IKMIS)
COP-NUR
COP-KEV-KRK
Fig. 4. Shear wave velocity distributions in the upper mantle. The heavy line represents a possible average velocity value. Da&xl lines give the depth uncertainty in the velocity distribution.
also down to depths of about 350 km, as pointed out also by Fuchs et al. (1983) and as is evident from the FENNOLORA experiment. Even though more deta2ed analysis is required to delineate the pattern of the heterogeneity, in-
version-of selected paths has opened up the possibility of finding homogeneous layering below about 350 km. fn the central part of the Baltic Shield this seems to be the depth to w&h the bottom of the shield might extend.
Dt.EP
STRUCTURE
OF FENNOSCANDtA
FROM
DISPERSION
OF RAYLEIGH
Acknowledgements
WAVES
149
Guggisberg,
B. and
velocity
I gratefully
acknowledge the financial support of the Italian Ministry of Education (MPI 40%).
and
model
its posstble
Knopoff. Geofis.,
N.N.
upper
and
mantle
of Raylergh
Knopoff, under
L.
1974. The structure
the United
waves.
Geophys.
States
of the
Sot,,
36: 51%
.I. and
structure
Dorman.
J.. 1963.
m the Canadian
Seismic
shield.
waves
and
Bull. Setsmol.
earth
Sot. Am..
H.. Prrhonen,
thickeness
S. and
Husebye.
m Fennoscandia.
Geophys.
A.L . Ptsarenko,
Calcagnile.
Tectonophysics,
G. and Panza.
structure
under
the dispersion
Sot..
of Rayleigh
m
mantle
Sea from
waves. Tectonophysics.
G. and
Scarpa,
R.. 1985
European-Medtterranean Tectonophysics,
47, 59-
higher
Eurasia.
area
from
H. and
mantle
Gtth,
structure
of the
seismological
Geophys.
data
data.
G., 1980. New mferences
in western
J R. Astron.
Europe
and
northern
hthosphere
beneath
of the
J. Geophys .
Fennoscandra.
D.. and
of the lower lithosphere
1975.
R.W . 1973. Ray&h shear
velocity
Univ. California.
wase phase velocities
structure
Los Angeles.
K., Kammski,
W. and
from the top of the mantle
of Fennoscandia
and upper Thests,
of setsmtc
Europe.
In. R. Maaz
transttion
Veroeff.
modes and the determmatron
Thesis.
of upper
Univ. Utrecht.
1977.
The
from
the disperston
upper
mantie
under
wes.em
Europe
of Rayleigh
modes.
J. Geo-
phys., 43, 265-285 Noiet.
G
and
Panza,
G.F.,
1976. Array
analysis
of seismtc
Pure Appl. Geophys
C., 1983 zone. Terra
Reflection Cogmta,
3: 103 B.. 1986.
Eme zweidimensionale
mtsche interpretation des oberen
Erdmantels
Fennolora).
refraktionssets-
Noponen.
I.. 1966. Surface
Panza,
G.F.
and
Scalera,
measurement. G.F..
features from Panza,
velocities
G.. 1978
in F nland
Higher
Pure Appl. Geophys.
Bull
Muller.
mode
St. and Calcagnde.
surface
waves
and
drspersion
116: 1283-1294 G., 1980a. The gross
of the lithosphere-asthenosphere
seisnnc
system m Europe
body
waves.
Pure
Appl.
118: 1209-1213.
G F , Calcagnde.
1980b
wave
Sot. Am.. 56: 1093- 1104.
La struttura
Sacks.
IS..
Snoke,
thickness
beneath
unter dem Fennoskandischen Ph D. Thesis, ETH Zimch.
Stuart,
G.W..
Sea regron
der Geschwindigkeits-Tiefen-Struktur Schild
,
114: 775-790
G.,
Scandone,
profonda
P. and
dell’area
Muller,
meditzrranea.
St. SCI-
A. and the
Husebye, Baltic
E.S.
Shield
1979. Lithosphere Tectonrtphysics.
56
101-110.
Guggrsberg,
(Prolekt
Physics.
enze, 141: 60-71.
Calif.
Prodehl,
Scandmavta
Dtsperston
Sohd-Earth
G..
Geophys.
46: 369-384. mantle
G.A . 1972.
An 1. Geophys.,
m northern
and northern
Inferred
Panza,
Sot.. 61: 459-478.
K.. 1979. Seismic investigatrons
Ann.
Phys. Erde. 31: 3999403
structure.
Setsmol.
A. and Nolet,
mode
B.R. and Fuchs,
subcrustal
Deep
118: 93-111.
M., Nercessian.
from
Fuchs.
of osctllation.
surface waves: hmits and posstbrhties.
Calcagnde.
Clymer.
Neunhofer,
Nolet,
1978. Crust and upper
man-
waves.
56. 111-122.
J. R. Astron.
71
Cassel,
Tectonophysics,
Seismology
90: 19-35.
G.F..
7:
of upper
of Rayleigh
the Blue Road profile
(Editor).
the Baltic shield and the Barents
Resolution
The fme structure
C.-E.. 1979
Nolet, G.. 1976. Higher
Fennoscandia.
Cara.
Lund,
Zentralmst.
system
1977 modes
analysts
E.S., 1980. Crustal
G . 1982. The lithosphere-asthenosphere
Cogmta.
V.F. and Progrebmsky,
On a frequency-time
waves m middle
63: 759-774 Calcagnde,
the Baltic shield
Terra
33: 491-505.
surface
53: 167-210. Bungum.
G.F.,
usrng higher
underneath
539. Bnme,
beneath
stgmficance.
28: 211-218.
from the dispersion
J. R. Astron.
tectonic
L. and Panza.
Levsbm, Biswas.
A.. 1987. A two dtmenstonal
631-638. tle structure
References
Berthelsen,
for the lithosphere
Astron.
1978. The upper from Rayletgh
Sac.. 52: 367-382.
mantle
structure
wave dispersion.
of the North Geophys.
J. R