EPSL

ELSEVIER Earth and Planetary Science Letters 171 (1999) 329-334

www.elsevier.com/locate/epsl

Express Letter

Space-geodetic estimation of the Nazca-South America Euler vector

D. A n g e r m a n n *, J. K l o t z , C. R e i g b e r
GeoForschungsZentrum Potsdam, Division 1, 14473 Potsdam, Germany
Received 10 February 1999; revised version received 8 June 1999; accepted 1 July 1999

Abstract

GPS data from four sites in the Nazca Plate (Easter Island, Galapagos, Robinson Crusoe and San Felix Islands) and
from five sites in the stable core of the South American Plate enabled us to estimate the Euler vector of the Nazca Plate
with respect to South America. The observed velocities of Easter Island (6.6 cm/yr at 102.3°), Galapagos (5.1 cm/yr at
90.0°), Robinson Crusoe (6.6 cm/yr at 80.1°) and San Felix (6.0 cm/yr at 82.1°) are significantly slower than the global
plate model NUVEL-1A velocites for those four sites. Our estimated Euler pole is located at 48.8°N, 91.7°W with a rate
of 0.59°/m.y. The NUVEL-1A and earlier space-geodetic studies give rotation rates that are 20% faster. © 1999 Elsevier
Science B.V. All rights reserved.

Keywords: Global Positioning System; movement; plate tectonics; Nazca Plate; South American Plate; plate convergence

1. Introduction most major plates (e.g. North America, Europe, Pa-

cific, Australia, South America), some plates, such
During the past decade, space-geodetic observa- as the Nazca Plate, are sparsely covered.
tions have become a key tool in plate tectonic stud- Pardo-Casas and Molnar [6] calculate by combin-
ies. In addition to VLBI, SLR, DORIS and PRARE ing reconstructions of the South American, African,
stations, a vast number of GPS sites commenced Antarctic, Pacific and Nazca plates a convergence
operation during the last years, leading to a fairly rate of 11.0-11.2 c m / y r between 10°S and 40°S
homogeneous distribution of tracking stations across since 26 Ma. They conclude, that a change in the
the globe. The data from these space-geodetic track- direction of relative motion between the Nazca and
ing sites provide an excellent basis for measuring South American plates was small but the rate of con-
plate motions over the time scale of years. Existing vergence has not been constant. The seismic slip rate
geophysical plate motion models, e.g. NUVEL-1A along the Chile trench estimated from the slip in the
[1] reflect plate motions averaged over millions of M = 9.5 1960 Chile earthquake gives a convergence
years. In the last years an increasing number of pa- in the range of 10.6 to 31 c m / y r [7] and [81. [9]
pers deal with the estimation of the present-day Eu- analyzed profiles across a number of ridge segments,
ler (angular velocity) vectors for the different plates and showed that spreading has slowed to an aver-
based on space-geodetic data (e.g. [2-51). Though age rate over the last three million years of about
space-geodetic data are fairly well distributed over 6.4 cm/yr. The Nazca and South American plates
were included in the plate motion models RM1 [10],
*Corresponding author. E-mail: dang@gfz-potsdam.deu Chase [11], RM2 [12], NUVEL-1 [13,14] and NU-

0012-821X/99/$ - see front matter © 1999 Elsevier Science B.V. All fights reserved.
PII: S001 2-821X(99)00173-9
330 D. Angermann et aL /Earth and Planetary Science Letters 171 (1999) 329-334

VEL-1A [1]. The predicted convergence is highest GALA is poorly estimated [16], we used the CODE
for RM1 (11.0 cm/yr at 40°S, 74°W) and decreases solution [17], which results from an observational
successively with more recent models. NUVEL-1A time span of 9 months for GALA.
predicts a convergence rate of 8.0 cm/yr at the loca- The GPS data was processed with the GFZ Soft-
tion of the 1960 Chilean earthquake. Space-geodetic ware EPOS (Earth Parameter and Orbit System) [ 18].
estimates of the present-day Enler vector for the The processing procedure of this software package is
Nazca Plate are given by [3,4] and [5]. The results based on undifferenced phase and pseudorange ob-
reported by these authors are based on the data from servations. The network computation is done in the
two stations on Easter Island and the Galapagos Is- form of daily solutions with ionosphere-free linear
lands, motion at the latter being derived from two combinations for the carrier phases and pseudor-
different site locations. Since a minimum of two anges, using an elevation cut-off angle of 15° and
sites is required, the uncertainties of the estimated a sampling rate of 4 rain. The combined IGS orbits
parameters (Euler pole and rate) for the Nazca Plate and the corresponding Earth Orientation Parameters
are large. [3] and [4] estimated a convergence rate (EOPs) [19] were fixed and all station coordinates
at 40°S, 74°W of 8.2 cm/yr and 7.8 cm/yr, respec- were solved for without constraints (fiducial-free
tively. These estimates are in close agreement with solution). The ambiguities resulting from the iono-
the NUVEL-1A prediction of 8.0 cm/yr. sphere-free linear combination of phases were ad-
This paper focusses on the estimation of the Euler justed as free parameters (float solution) along with
vector for the Nazca plate with respect to a stable the receiver and satellite clock parameters for each
South America. The solution is based on four GPS observation epoch and the tropospheric parameters
stations on the Nazca Plate, viz. the IGS stations for each station (estimated at 4-h intervals). The
located on Easter Island and the Galapagos Islands, SAGA data were processed simultaneously with data
and the stations on Robinson Crusoe and San Fe- from the IGS station network to transform the SAGA
lix Islands (Chile), occupied periodically as part of network into the ITRF96 frame. Global network so-
GFZ's SAGA (South American Geodynamic Activ- lutions were computed using about 20 globally dis-
ities) network in the Central and Southern Andes tributed IGS stations, whereas regional SAGA solu-
[15]. tions used only IGS stations in South America [20].
Two methods were used to assess the quality
of the SAGA network coordinates. (1) The daily
2. GPS data and processing repeatability estimated using the residuals of a
Helmert-transformation between mean and daily net-
The two SAGA sites on the Nazca plate were work solutions as the assessment criteria. The mean
observed simultaneously with about 100 SAGA sta- station coordinates residuals are of the order of 1-3
tions during two large GPS campaigns in March mm in the horizontal and 5 to 6 mm in the ver-
1994 and November/December 1996. In 1994 both tical direction. (2) The network accuracy estimated
island stations were occupied for two weeks. For lo- by comparing the SAGA network solutions with the
gistical reasons only the station on Robinson Crusoe ITRF96 station coordinate solution. The means of
Island could be observed during the main SAGA' 96 the residuals after a Helmert-transformation between
campaign. Fortunately though, the station on San the global SAGA 94/96 solutions and the ITRF96
Felix could be reoccupied for a three-day session station coordinates (19 IGS stations) are 1 cm hor-
two months after the main campaign. The IGS sta- izontally and 2 cm in the vertical direction. The
tion Easter Island was not operational during the regional network accuracy with respect to the South
SAGA'94 campaign, so for this station the ITRF96 American reference frame is assessed by comparing
station coordinates and velocities based on GPS and the regional SAGA solutions with the 1TRF96 coor-
SLR data were used. The GPS station Galapagos dinates of the IGS stations in South America. The
became an IGS station at the beginning of 1996, accuracy of the regional SAGA solutions is of the
and only few observations exist for the years 1996 order of 2-5 mm horizontally and better than 1 cm
and 1997. Since the ITRF96 velocity for the station vertically.
 D. Angermann et al./Earth and Planetary Science Letters 171 (1999) 329-334 331

3. Reference frame and site velocities Table 1

The site velocities with respect to the stable part of South
America along with their one sigma standard deviations
As described above, the SAGA solutions were
computed using fixed combined IGS orbits and Stations North East Height
EOP's. The station coordinates were solved for with- (mm/yr) (mm/yr) (mm/yr)
out constraints. Hence the reference frame of these KOUR 0.1 4- 0.8 0.5 4- 2.6 0.4 4- 2.7
fiducial-free solutions is defined by the IGS orbits FORT 0.7 4- 0.7 -0.5 4- 1.6 2.2 -t- 1.6
and the EOP's. In a second step the S A G A solutions BRAZ -2.3 4- 2.6 -3.7 -4- 5.1 -10.1 4- 5.0
LPGS -0.9 -4- 1.6 0.6 -4- 2.1 1.1 4- 1.7
were transformed to the ITRF96 station coordinates
ASC1 -0.9 4- 2.7 3.0 :t: 5.0 2.1 4- 2.0
by performing a Helmert-transformation. As the fo-
cus in this study is on relative plate motions, a BOGT 5.2 -4- 1.5 9.9 4- 3.1 0.1 4- 1.5
AREQ 1.3 4- 0.6 11.7 4- 0.8 -1.2 4- 0.8
regional Helmert-transformation was applied, which
SANT 2.5 -t- 1.5 17.8 4- 1.4 8.1 4- 1.7
leads to significant smaller residuals for the South
American IGS stations when compared with those EISL -14.0 4- 1.1 64.6 4- 1.0 1.8 4- 1.4
IRCR 11.4 4- 1.8 65.0 4- 1.9 -8.1 4- 3.7
derived from a global transformation (see [20]). ISFE 8.2 4- 2.2 59.5 4- 2.5 1.4 ± 3.8
Site velocities can be related to either a global GALA 0.0 4- 3.1 51.3 -4- 6.8 0.0 4- 2.4
or a regional reference frame. The relative velocities
The five stations in the upper part of the table belong to the
in a regional frame are more precise, as a result of
stable part of the South American Plate. The velocities of the
spatial correlations of the estimated parameters. The stations Robinson Crusoe Island (IRCR) and San Felix Island
South American reference frame used here assumes (ISFE) are estimations from the SAGA94/96 campaigns; the
that the IGS stations Fortaleza (FORT), Kourou other stations show the ITRF96 velocities. The ITRF96 velocity
(KOUR), Brazilia (BRAZ), La Plata (LPGS) and of station GALA has been replaced by the CODE-solution [17].
Ascension Island (ASC1) represent the stable body
of South America. The ITRF96 velocities for these
sites result in an absolute Euler pole position (in the for Galapagos, were transformed to the datum of
ITRF96 frame) of - 6 . 7 4-2.8°N, - 1 4 7 . 0 4- 10.7°E stable South America, making sure that the different
with a rotation rate of 0.10 4- 0.01°/m.y. The cor- solutions are related to the same reference frame.
responding values of the global plate motion model Fig. 1 shows the horizontal station velocities for the
(NNR) N U V E L - 1 A [1,13] o f - 2 5 . 4 ° N , - 1 2 4 . 6 ° E eight IGS stations in South America and the four
and 0.12°/Myr differ significantly from the space- stations of the Nazca Plate along with their 95%
geodetic estimates (probably due to the use of a confidence ellipses. The SAGA 94/96 velocities and
different datum). The internal consistency of the the corresponding ITRF96 velocities agree within
reference frame applied here is best assessed by ex- 1-3 mm/yr, the differences being insignificant, as
amining the residuals of all stations located within indicated by their 95% error ellipses. Table 1 shows
the stable part of South America. The residuals of the the station velocities along with their standard devi-
ITRF96 velocities with respect to the ITRF96-based ations. The stations Bogota, Arequipa and Santiago,
Euler vector are shown in Table 1. The horizontal ve- which are located within the plate boundary zone,
locity residuals for the five stations of the stable part show significant motions of 1 - 2 c m / y r with respect
of South America (upper part of Table 1) are in the to the South American reference frame. For the sta-
range of 1-3 m m / y r and lie below their one sigma tions on the Nazca Plate, the observed velocities are
standard deviations. The vertical velocity component significantly smaller than the N U V E L - 1 A predic-
of the station Brazilia (1 cm/yr) is not significant tions. Kendrick et al., 1999 [21] realized in a recent
at the 95% confidence level. These results indicate paper a reference frame that represents the rigid core
that the five stations selected represent stable South of the South American continent by minimising the
America. motions of those stations located within the stable
Using the ITRF96-based Euler vector the SAGA part of South America. The station velocities given
velocities, including the GPS-derived velocities of by [21] agree within their standard deviations with
the IGS sites derived at GFZ and the C O D E solution the results derived here.
332 D. Angermann et al. / Earth and Planetar), Science Letters 171 (1999) 329-334

260 ° 280 ° 300 ° 320 =

0° 0o

.20 ° -20

°,40°¸ ~40 °

260 ° 280 ° 300 ° 320 °

Fig. 1. ITP~96 and SAGA-derived horizontal displacements with respect to stable South America along with their 95% confidence
ellipses, including the NUVEL-IA velocities for the stations located on the Nazca Plate. The ][TRF96 velocity of station GALA is poorly
estimated, therefore the CODE-solution [10] is used for this station.

4. Nazca-South America Euler vector and tot (pole and rotation rate) for the N a z c a Plate with
discussion respect to the South A m e r i c a n Plate in a least squares
adjustment, T h e results o f the e s t i m a t i o n are s h o w n
T h e horizontal relative v e l o c i t i e s o f the four in Table 2, along w i t h the c o r r e s p o n d i n g N U V E L - 1 A
stations E a s t e r Island (EISL), G a l a p a g o s ( G A L A ) , values and the G P S results o f [3] and [4]. T h e esti-
R o b i n s o n Crusoe ( I R C R ) and San F e l i x Island m a r i o n o f the rotation rate (0.59°/m,y.) d e r i v e d here
(ISFE) w e r e used to e s t i m a t e the relative E u l e r vec- differs significantly f r o m the N U V E L - 1 A prediction

Table 2
The Nazca-South America Euler vector (pole ~md rotation rate) obtained from different sources

Source ][,at. Long, 6~) O'max ~min Azim cfw

(~'N) (°W) (°/m.y,l (°) (°) (°) (°)
Final solution 48.8 91.7 0.59 4.0 1,9 -3.3 0.014
Three sites solution 45~4 92.3 0.60 6.2 2.2 -3.5 0.020
Two sites solution 45.8 92.6 0.61 4.1 1.6 2.9 0.0~2
NUVEL1--A 55.9 94,0 0.72 3.7 1.5 i0 0.02
Larson et al. [3] 43.8 84.8 0,74 9.1 5.5 18 0.07
Norabuena et aL [4] 40.6 90.5 0.71 4.4 2.2 -7.4 0,03
One sigma error ellipses are specified by the angular lengths of the principal axes and by the azimuths of the major axis, Rotation rates
are in degree per millions of years.
 D. Angermann et al./Earth and Planetary Science Letters 171 (1999) 329-334 333

.120 ° .90 ° .60 ° we estimated the Euler vector using the two stations
60 ° 50 ° Easter Island and Robinson Crusoe. The resulting
Euler vector differs by 3.0 ° in latitude, 0.9 ° in lon-
gitude and 0.02°/m.y. in rate from the final 4 sites
solution including the 4 sites. The small insignif-
icant differences between the 2 site, 3 site and 4
site solutions (see Table 2) demonstrate that the re-
300 30 ° ported results are stable. The Euler vector computed
in this study suggests a 17% slower relative motion
between the N a z c a and the South A m e r i c a n plates
than predicted by N U V E L - 1 A .

0o 0o

Acknowledgements

This work was supported by the ' G e o -

ForschungsZentrum Potsdam' and by the SFB 267
'Defomaationsprozesse in den A M e n ' . Thanks are
-30' -30 °
expressed to the Instituto Geografico Militar (IGM)
Chile for supporting the GPS observations in Chile
as well as to the IGS and the IERS communities
for providing the GPS data, IGS products, ITRF96
.120 ° .90 ° .60 °
station coordinates and velocities. Michael Bevis,
Fig. 2. Our GPS-derived Euler vector for the Nazca Plate with
Eric Kendrick and Bob Smalley provided us with
respect to stable South America along with the 95% confidence
ellipse is shown in comparison to the (NNR)NUVEL-1A model preprints and many fruitful discussions. [ R O ]
and the results of the earlier space-geodetic studies. The pre-
dicted motions of these different studies are shown for four sites
on the Nazca Plate. References

of 0.72°/m.y. and from the estimates of [3] and [4] [1] C. DeMets, R.G. Gordon, D.E Argus, S. Stein, Effect of
recent revisions on to the geomagnetic reversal time scale
(see Table 2 and Fig. 2). The residual motions of
on estimates of current plate motions, Geophys. Res. Lett.
the four stations on the N a z c a Plate with respect to 21 (1994) 2191-2194.
the Euler vector estimated in this paper average 2 [2] G. Blewitt, R.S. Kawar, EB.H. Davies, Globally consistent
ram/yr. These values are not significant, indicating rigid plate motion: fiducial-free Euler vector estimation,
that these stations represent the N a z c a Plate quite Proc. IGS Analysis Center Workshop, Darmstadt, 1998,
191-202.
well.
[3] K. Larson, J.T. Freymueller, S. Philipsen, Global plate
In order to verify the Euler vector estimation, velocities from the Global Positioning System, J. Geophys.
solutions were computed to include only 3 and 2 Res. 102 (1997) 9961-9981.
sites. (1) Station Galapagos was excluded due to the [4] E. Norabuena, L. Leffter-Griffin, A. Mao, T. Dixon, S.
short observation time span (9 months) in the C O D E Stein, I.S. Sacks, L. Ocola, M. Ellis, Space geodetic obser-
vation of Nazca-South America convergence across Central
solution and the Euler vector was re-estimated using
Andes, Science 279 (1998) 358-362.
the horizontal velocities of Easter Island, San Felix [5] E Sillard, Z. Altamimi, C. Boucher, The ITRF96 realization
and Robinson Crusoe. The differences between the and its associated velocity field, Geophys. Res. Lett. 25
resulting Euler pole and the final 4 sites solution (1998) 3223-3226.
are 3.4 ° in latitude, 0.6 ° in longitude. The rotation [6] E Pardo-Casas, R Molnar, Relative Motion of the Nazca
(Farallon) and South American plates since late Cretaceous
rate agrees to within 0.01°/m.y. (2) Station San Felix time, Tectonics 6 (1987) 233-248.
was also excluded, on the basis of the less precise [7] H. Kanamori, Seismic and aseismic slip along subduction
solution taken with the data in D e c e m b e r 1996, and zones and their tectonic implications, in: M. Talwani, W.C.
334 D. Angermann et al./Earth and Planetary Science Letters 171 (1999) 329-334

Pitman, III (Eds.), Island Arcs, Deep-sea Trenches mad J. Reinking, J. Viramonte, R. Perdomo, V.H. Rios, S. Bari-
Back-Arc Basins, Maurice Ewing Ser. 1, AGU, Washington entos, R. Barriga, O. Cifuentes, GPS-derived deformation
D.C., 1977, pp. 163-174. of the Central Andes including the 1995 Antofagasta M~ =
[8] S. Stein, J.F. Engeln, C. DeMets, R.G. Gordon, D. Woods, 8.0 earthquake, Pure Appl. Geophys. (1999) (in press).
P. Lundgren, D. Argus, C. Stein, D.A. Wiens, The Nazca- [16] C. Boucher, Z. Altarnimi, E Sillard, Results and Analysis
South America convergence rate and the recurrence of the of the ITRF96, IERS Tech. Note 24 (1998) 166 pp.
great 1960 Chilean earthquake, Geophys. Res. Lett. 13 [17] M. Rothacher, G. Beutler, T.A. Springer, S. Schaer, IGS-
(1986) 713-716. Analysis Center Solution SSC(CODE)98P01, University of
[9] E.M. Herron, S.C. Cande, B.R. Hall, An active spreading Berne, Berne, 1998.
center collides with a subduction zone: a geophysical sur- {18] D. Angermann, G. Baustert, R. Galas, S.Y. Zhu, EPOS.P.V3
vey of the Chile margin triple junction, Geol. Soc. Am. (Earth Parameter and Orbit System): Software User Man-
Mem. 154 (1981) 683-702. ual for GPS Data Processing, Scientific Technical Report
[10] J.B. Minster, T.H. Jordan, P. Molnar, E. Haines, Numerical STR97/14, GeoForschungsZentrum, Potsdam, 1997, 52 pp.
modeling of instantaneous plate tectonics, Geophys. J.R. [19] International GPS Service for Geodynamics, in: J.F. Zum-
Astron. Soc. 36 (1974) 541-576. berge, D.E. Fulton, R.E. Neilan (Eds.), 1996 Annual Re-
[11] C.G. Chase, Plate kinematics: the Americas, East Africa, port, IGS Central Bureau, Jet Propulsion Laboratory, Cal-
and the rest of the world, Earth Planet. Sci. Lett. 37 (1978) ifornia Institute of Technology, Pasadena, CA, 1997, 446
355-368. PP.
[12] J.B. Minster, T.H. Jordan, Present-day plate motions, J. [20] D. Angermann, J. Klotz, C. Reigber, Geodetic datum defi-
Geophys. Res. 83 (1978) 5331-5354. nition of the SAGA Network, Proc. IAG Section II Sympo-
[13] D.E Argus, R.G. Gordon, No-net-rotation model of cur- sium, Munich, 1999 (in press).
rent plate velocities incorporating plate motion model NU- [21] E. Kendrick, M. Bevis, R. Smalley, O. Cifuentes, E Galban,
VEL-1, Geophys. Res. Lett. 18 (1991) 2038-2042. Current rates of convergence across the Central Andes: es-
[14] C. DeMets, R.G. Gordon, D.F. Argus, S. Stein, Current timates from continuous GPS observations, Geophys. Res.
plate motions, Geophys. J. Int. 101 (1990) 425-478. Lett. 26 (1999) 541-544.
[15] J. Klotz, D. Angermann, G.W. Michel, R. Porth, C. Reigbel,