Data processing

The solution presented here is a combination of two different solutions performed 
independently with different softwares (GAMIT / GLOBK and GIPSY-OASIS II). The data 
sets are also different, having only 60% of the sites in common, and different measurements 
history at some sites. The main difference in processing strategies is that GAMIT explicitly 
forms double differences and solves for all baselines in a given network, while GIPSY uses 
the precise point positioning (PPP) strategy. The second important difference is the strategy 
used to estimate velocities. GLOBK combines all measurements to estimate positions and 
velocities in an unconstrained reference frame (the so called free network solution) and map 
them into a given reference frame (here ITRF2000) in a second stage. GIPSY estimates 
positions in a given reference frame (again ITRF2000) at every epoch, and velocities are then 
estimated by linear trend fitting of the position time series.

1.	Details on the GAMIT/GLOBK processing

The solution computed using the GAMIT / GLOBK software [Herring, 1999; King, 1999] 
includes more than 150 stations spread over Southern Asia spanning 11 years of data 
processed simultaneously. The data located in SE-Asia are GEODYSSEA (GEODYnamics of 
South and Southeast Asia) project) (1994-1996-1998) [Michel et al., 2001], APRGP (Asia 
Pacific Regional Geodetic Project) (1997-1998-1999-2000) [Govind, et al., 1999], THAICA 
(first order network of Thailand established by the Royal Thai Survey Department) (1994-
1996-2000), Myanmar campaigns performed in 1998-2000  [Vigny, et al., 2003] and Sulawesi 
(ENS/DEOS densification of the GEODYSSEA network in Sulawesi, Indonesia) (1999-2000-
2001-2002) campaigns [Walpersdorf et al., 1999; Simons et al., 2000; Vigny et al., 2002; 
Socquet et al., 2006]. Data located on the Indian plate include the West Nepal network 
measured in 1991-1995-1997-1998-2000 within the CIRES and IDYLHIM projects [Bilham, 
et al., 1997; Larson, et al., 1999; Jouanne, et al., 1999], the Kathmandu transect measured in 
1998-1999-2000 by the LDG-CEA (Commissariat  lEnergie Atomique) [Avouac, et al., 
2001] and HYDE station (Hyderabad) measured by the National Geophysical Research 
Institute in Hyderabad (NGRI, India) and the Geodaetisches Institut der Universitaet Bonn 
(Germany) in 1996-1997-1998-1999-2000.. Finally, all IGS stations available in the area at 
the time of any of our measurements are included 
Twenty-four-hour measurement sessions are reduced to daily positions. For every daily 
session theoretical values for phase and pseudorange observables are modelled. The model 
parameters such as stations coordinates, phase ambiguities and tropospheric zenith delay (one 
every 3 hours) are adjusted by least squares method. The observations are sampled every 30 
seconds with an elevation cut-off angle of 15 degrees. We use the "LC" or "L3" Ionosphere 
free combination of the two-phase data and explicitly form the double differences (between 
every pairs of satellites and stations). During a first step the integer ambiguities are allowed to 
take decimal values. They are assigned to integer values in a second step, using the method 
developed by [Dong and Bock, 1989]. We also used IGS tables for modelling of antenna 
phase centre variations. Finally, we used precise orbits from the International GPS Service for 
Geodynamics (IGS) [Beutler, et al., 1993] that we did not re-estimate, except for the West 
Nepal data of 1991 where we used the orbits computed by [Larson, et al., 1997]. IGS orbits 
are fixed in different reference frames as time passes. This is taken into account by simply 
estimating the translation and rotation parameters of the reference frame at each epoch when 
combining the epoch solutions to estimate linear velocities at all sites. This is a standard 
procedure which gives similar results to the "relax" orbit kind of solutions.

One of the ways to assess the internal quality of a solution is to estimate daily repeatabilities, 
i.e. the scatter of the baseline measurements about their campaign average. This is done for 
every campaign and gives a higher bound of its noise level, provided that sufficient repetition 
generates significant repeatabilities. Baseline component repeatability values for each 
campaign are summarized in Table S1 for fixed ambiguities. Repeatabilities vary for each 
campaign, depending on the quality of measurements, the quality and homogeneity in the type 
of antennae used, the mean length of baselines for the considered campaign (the dispersion 
increases with the length of the baselines essentially due to uncertainties in satellites orbits) 
and the number of stations processed simultaneously.  Values range between 1 and 4 mm for 
the north component and between 3 and 8 mm for the east component, whereby more recent 
data generally show better repeatabilities.

The solution vectors and associated variance/covariance matrices of the independent daily 
solutions have been combined by GLOBK, based on a Kalman filtering approach, to obtain a 
multi-session free-network solution for each campaign with loose constrains on positions and 
velocities. Long term repeatabilities (i.e. the scatter of independent campaign determinations 
about their detrended average) have been performed for this solution and average values are 
about1.8 mm for the north component and 2.9 mm for the east component for baselines of 
about 1000 km (Fig S1). Velocities are not estimated at this stage but rather after the 
combination with the GIPSY solution.

2.	Details on GIPSY processing

The solution computed with the GIPSY-OASIS II v2.6.1 [Blewitt, et al., 1988] software uses 
a data set spanning 1994-2002. Again, GEODYSSEA (1994-1996-1998), THAICA (1994-
1996-2000) and Sulawesi (1999-2002) campaigns have been re-processed independently from 
the GAMIT solution. Theses are the data common to the GAMIT/GLOBK solution, and 
represent 60% of the sites used in the final combination. Three years of the permanent 
Malaysian (MASS) network (1999-2000-2001), Sulawesi (1996-1997-1998) and THAICA 
(2001) data have been added. Also, up to 30 IGS stations world wide spread have been 
included in the processing to allow an accurate mapping in the ITRF. 
The precise point positioning (PPP) strategy has been chosen, using fixed satellite orbits and 
clocks distributed by JPL (Jet Propulsion Laboratory). As in the GAMIT processing, the 
ionosphere-free linear combination of GPS phase and pseudorange data is used, here with a 
sample rate of 5 minutes and an elevation cut-off angle of 15 degrees. Tropospheric delay 
parameters (zenith path delay and gradients) are also estimated. Ocean loading parameters are 
included for each GPS location [Scherneck, 1991], and to account for the different GPS 
antennae, the NGS relative antenna phase centre corrections [Mader, 1998] are applied.     
Each campaign solution is projected on ITRF-2000 using up to 25 IGS stations. Then, 
velocities are obtained by fitting a linear trend through stations position by a mean square 
minimization. At that stage, no adjustments of the station positions are allowed. Outliers 
(positions differing by more than 2mm from their estimated value) are removed from the 
linear trend estimation. Those can be due to measurements errors of any kind or co-seismic 
induced displacements for example. A detailed description on the GIPSY solution and 
discussion of its uncertainties are given in [Simons, et al., submitted].

3.	Combination

A combined solution was made from the GAMIT and GIPSY processing. It spans the entire 
region from India and Nepal up to the south-eastern tip of the Indonesian archipelago. This 
solution was generated by combining loosely constrained variance-covariance matrices from 
the two independent solutions with the GLOBK Kalman filter. Common sites (the regional 
IGS stations we use and 60% of our campaign sites) enable a strong unification of the two 
solutions. The final mapping in the International Terrestrial Reference Frame (ITRF) 2000 is 
done by minimizing (in a mean square procedure) the departure of a selected list of IGS 
stations positions and velocities from their a priori values in the ITRF2000  [Altamimi, et al., 
2002]. On these stations, the velocity misfits range between 0 and 2 mm/yr, indicating an 
accurate mapping in the reference frame (Table S2). A comparison of velocities determined 
independently (GAMIT/GLOBK vs. GIPSY) at common sites is also made (Table S3). They 
differ by no more than 2 mm/yr at most sites, and these differences should probably be 
attributed to the usage of different orbits (IGS vs. JPL) and the slightly different antenna 
phase centre corrections (IGS vs. NGS) that were applied. Velocities in ITRF2000 of most 
sites involved in the combined solution with their 1-sigma uncertainties are given in Table S4. 
The velocities are plotted with respect to the Sunda block on Fig. 2 in the paper. 


References:
Altamimi, Z., et al. (2002), ITRF2000: A new release of the International Terrestrial Reference frame for earth science applications, Journal of Geophysical Research-Solid Earth, 107, art. no.-2214.
Avouac, J. P., et al. (2001), Seismic cycle in the Himalayas, Comptes Rendus Acad. Sci. Ser II-A, 333, 513-529.
Beutler, G., et al. (1993), Combining the orbits of IGS processing centers, in Proceedings of IGS analysis center workshop, edited by K. J., pp. 20-56.
Bilham, R., et al. (1997), GPS measurements of present-day convergence across the Nepal Himalaya, Nature, 386, 61-64.
Blewitt, G., et al. (1988), {GPS} geodesy with centimeter accuracy, in Lecture Notes in Earth Sciences, edited by E. G. a. R. Strauss, Springer-Verlag, New York.
Chamot-Rooke, N., and X. Le Pichon (1999), GPS determined eastward Sundaland motion with respect to Eurasia confirmed by earthquakes slip vectors at Sunda and Philippine trenches, Earth Planet. Sci. Lett., 173, 439-455.
Dong, D. N., and Y. Bock (1989), Global Positioning System Network Analysis With Phase Ambiguity Resolution Applied to Crustal Deformation Studies in California, Journal of Geophysical Research-Solid Earth and Planets, 94, 3949-3966.
Govind, R., et al. (1999), Combination of high precision space geodetic techniques: The Asia and Pacific Regional Geodetic Project 1997, in Satellite Dynamics, Orbit Analysis and Combination of Space Techniques, edited, pp. 797-807.
Herring, T. A. (1999), Documentation for the GLOBK software version 5.01, Mass. Inst. of Technol., Cambridge.
Jouanne, F., et al. (1999), Oblique convergence in the Himalayas of western Nepal deduced from preliminary results of GPS measurements, Geophys. Res. Lett., 26, 1933-1936.
King, R. W., and Y. Bock (1999), Documentation for the GAMIT GPS software analysis version 9.9, Mass. Inst. of Technol., Cambridge.
Larson, K. M., et al. (1999), Kinematics of the India-Eurasia collision zone from GPS measurements, Journal of Geophysical Research-Solid Earth, 104, 1077-1093.
Larson, K. M., et al. (1997), Global plate velocities from the Global Positioning System, Journal of Geophysical Research-Solid Earth, 102, 9961-9981.
Mader, G. L. (1998), GPS Antenna Calibration at the National Geodetic Survey, Tecnical report available at NGS website, National Geodetic Survey, Silver Spring, US.
Michel, G. W., et al. (2001), Crustal motion and block behaviour in SE-Asia from GPS measurements, Earth Planet. Sci. Lett., 187, 239-244.
Scherneck, H.-G. (1991), A parametrized solid Earth tide mode and ocean loading effects for global geodetic base-line measurements, Geophysical Journal International, 106, 677-694.
Simons, W. J. F., et al. (1999), Observing plate motions in Southeast Asia; geodetic results of the GEODYSSEA Project, Geophys. Res. Lett., 26, 2081-2084.
Simons, W. J. F., et al. (submitted), A Decade of GPS Measurements in S.E. Asia: (Re)Defining Sundaland and it's Boundaries, J. Geophys. Res.
Simons, W. J. F., et al. (2000), Geodynamics of S.E. Asia; first results of the Sulawesi 1998 GPS campaign; IAG general assembly, IAG general assembly, 121, 271-277.
Socquet, A., et al. (2006), Micro block rotations and fault coupling in SE Asia triple junction (Sulawesi, Indonesia) from GPS and Earthquake Slip Vector data, Journal of Geophysical Research-Solid Earth, submitted.
Vigny, C., et al. (2002), Migration of seismicity and earthquake interactions monitored by GPS in SE Asia triple junction: Sulawesi, Indonesia, Journal of Geophysical Research-Solid Earth, 107, art. no.-2231.
Vigny, C., et al. (2003), Present day crustal deformation around Sagaing fault, Myanmar, Journal of Geophysical Research-Solid Earth, 108(B11), 2533.
Walpersdorf, A., et al. (1998), GPS compared to long-term geologic motion of the north arm of Sulawesi, Earth Planet. Sci. Lett., 159, 47-55.
Walpersdorf, A., et al. (1998), Determining the Sula block kinematics in the triple junction area in Indonesia by GPS, Geophysical Journal International, 135, 351-361.


