Comparison with previous surveys

We combined the daily solutions of the October 2003 GPS survey with daily solutions from the 1999 and 2001 surveys. Figure 6.3 shows the position time series for the campaign sites observed during the October 2003 survey. The (usually) linear change in position from one survey to the next corresponds to the absolute motion of the sites, since the coordinates are calculated in ITRF2000, a dynamic reference frame. A coseismic offset would appear as a break in that linear trend between the 2001 and 2003 surveys. From that figure, the following observations can be made:

• ARRO: linear trend, no clear break in the time series.
• CAMP: linear trend, no clear break in the time series. Note the large scatter in 1999 and 2001, in particular on the vertical component. This is probably due to a satellite dish that was located next to the site, but was removed in 2003. For that same reason, the vertical component is probably not very reliable in 1999 and 2001.
• CAPT: linear trend, no clear break in the time series.
• CASX: linear trend, no clear break in the time series.
• CRIS: no clear break in the time series. Note that we had significant problems at that site in 1999 and 2001. At the end of the 1999 survey, we found that the tripod was located about 13.6 cm to the SE of the benchmark. The tripod had obviously been tampered with, but we were not able to find exactly when this happened. This explains the scatter of the 1999 positions and the offset (in particular in the east component) compared to the next surveys. At the end of the 2001 survey, we discovered that the electrical power provided by the military had been disconnected after less than 24 hours of operations. We therefore have only one short measurement session for that survey.
• HIDA: there seem to be a break in the linear trend of the horizontal position time series between 2001 and 2003, possibly a result of coseismic displacement of the site.
• LAVE: linear trend for the north component. There seem to be a break in the time series for the east component.
• MONC: linear trend, no clear break in the time series. Note that the antenna was pointed to the south (instead of north) during the entire 1999 survey. This does not seem to have perturbed the results.
• PEPE: linear trend for the east component. The north component shows a significant offset between the 1999 and 2001 surveys. The 1999-2001 trend is expected to be positive for that site, as can be seen for all other sites. This points to a problem during either the 1999 or 2001 survey. We had not identified this problem before, we are currently investigating its source.
• REUN: linear trend in the north time series. Clear offset in the east and vertical time series. This offset is likely to be due to coseismic effects.
• SAGO: linear trend, no clear break in the time series.
• TERR: linear trend, no clear break in the time series.

In summary, only sites REUN and HIDA show a clear break in the position time series that can be interpreted as a result of a coseismic displacement. Site CRIS and PEPE cannot be used for estimating coseismic displacements because of data quality issues during the 1999 and/or 2001 surveys.

Interestingly, we find that the vertical component has a negative long term-trend at many GPS sites in the northern Dominican Republic. It is however too early to tell whether this pattern results from tectonic loading, sediment compaction, ground water migration, or is a geodetic artifact.

Figure 6.3: Position time series for the 12 campaign GPS sites observed during the October 2003 survey. Each point represents a position derived from a 24-hour solution, with an error bar representing its formal uncertainty. Top panels: NS component; middle panels: EW component; bottom panels: vertical component. The vertical dashed line indicates the day of the earthquake.

Figure 6.4: .

Figure 6.5: .