We tested the sensitivity of the pre-2005 vertical motions by recursively removing from the time series one pre-earthquake measurement at a time and re-fitting the linear trend to the rest. At both CHON and BANH, removing any single measurement from the height time series of BANH does not corrupt the uplift trend. Further, the BANH height time series is very consistent with that of CHON. If measurements in Nov-1998 and Oct-2000,
are removed from CHON time series, its up–down pattern is the same as depicted by that of BANH. This similarity is not repeated after the earthquake, when nonsteady motion is expected even though measurements and processing were carried out in the same fashion as in the pre-earthquake period. This suggests that the pre-2005 similarity between CHON and BANH did not result from any systematic effects of GPS, but rather from the tectonic signal. Altogether, this confirms the presence of preearthquake uplift that was captured by GPS despite the noisy appearance of the data.