In this section we calculate the velocities, assuming REYK is fixed, using data from the beginning of measurements until December 31, 2001. The offsets due to the radome installation are removed from the vertical components of the data assuming an offset of -20 mm at the times of radome installation and an offset of 20 mm when the radome is removed. The offsets due to the SISZ June 2000 earthquakes are kept in the data and the formal errors are scaled by 4.0 in east and north and 2.5 in vertical before the velocity estimation. We note that since the coseismic displacements have not been removed from the data before velocity estimation the velocities at the stations including the offsets (HLID, HOFN, HVER, HVOL, OLKE, SOHO, THEY and VOGS) are not neccesarily representing the average plate velocities. Furthermore, not all stations were recording at the time of the earthquake sequence so the velocities have different physical interpretation depending upon if the station was recording data or not (see Figure 21).
The results are shown in Table 6. To visualize the data in Table 6, a constant velocity vector representing the half-spread NUVEL-1A velocity [DeMets et al. (1990)], [DeMets et al. (1994)], at REYK, vr = 9.6 mm/yr west and 2.1 mm/yr north, is added to the velocity vectors in Table 6. The vector vr is obtained from the NUVEL-1A NNR model [DeMets et al. (1994)] by assuming that the plate velocities at REYK and HOFN are equal, but in opposite directions. This modification emphasizes the rifting in Iceland showing relative plate motions, i.e. on which tectonic plate the stations are located. The modified velocities are shown in Figure 22. Figure 23 is a close-up of the Hengill area.
Figure 22 shows quite well the plate spreading
although the earthquake offsets are in the data for stations
HLID, HOFN, HVER, HVOL, OLKE, SOHO, THEY and VOGS. The stations
moving towards WNW (REYK, OLKE, SKRO and AKUR) are on the
North-American plate whereas the other stations moving towards
ESE or SE (HLID, VOGS, VMEY, THEY, SOHO, HVOL, HOFN and RHOF)
are on the Eurasian plate.
The chi-squared values in Table 6
vary highly. Large
values are observed, up to
for the east
component of VOGS. This is to be expected since a straight line
is a poor model for time series that include the offsets due
to the June 2000 earthquakes (stations HOFN, HVER, HVOL, OLKE, SOHO, THEY and VOGS).
This is also reflected in the WSTD values in Table 6.
Vertical movements will be discussed in Section 4.2.5.