D27 Estimated daily positions, SW Iceland CGPS network, for current volcanic deformation – IMO

There are presently 17 CGPS sites in Iceland, all of which are included in routine processing of ISGPS data. The ISGPS network now covers not only the SW part of Iceland, but follows roughly the plate boundary to North Iceland (Figure 1).

We estimate the daily positions of all stations in the network using the Bernese V4.2 software. The positions are tied to the ITRF97 reference frame by tightly constraining the coordinates of the reference station (REYK) to its ITRF97 coordinates.

We have estimated the daily positions for all CGPS sites in Iceland from March 18, 1999 to December 31, 2003. We view the results as time series (Figure 2), showing the movements of the stations relative to REYK in east, north and vertical coordinate components. We can fit the time series with a simple model:

y = a + bt + Acos(ωt + ψ) + Σg _jH(t_j)

which consists of a straight line, an annual term (ω=2 π) and a Heaviside function term to get offsets due to earthquakes or instrumental changes. From the rate (parameter b) we get the long-term average motion of the stations, shown in figure 2. Generally the movements of the stations agree well with plate motion models (e.g. NUVEL-1A) but discrepencies are observed close to the plate boundary or due to volcanic deformation. The only area currently showing detectable volcanic deformation on the ISGPS network is Katla, which is long due to erupt. The stations closest to Katla are moving outwards from the central caldera indicating increasing pressure in the magma chamber. This is further supported with campaign GPS measurements at nunataks on Katla. It is impossible to predict when the volcano will erupt from the data, but if sudden increases are observed in inflation rate, we can expect an eruption to occur within days.

IMO has worked further on completing deliverable D26: “Methods for frequent analysis” which was not completed at M12. We have now obtained results showing that processing the data in 1 hour batches is useful for near real-time monitoring of the crust. We chose two test time periods for the analysis: 1) the first 2 weeks of 2003 and b) 2.5 weeks covering the June 2000 earthquake sequence (Figure 2). The former time period was used for inspection of the statistical properties of the results and the latter period was used to test if we could identify rapidly coseismic deformation. We processed the data in 1, 2 and 3 hour batches and used 2 existing processing strategies. The results immediately indicated that one of the strategies yielded by far more stable results, but both strategies (1 hour batch processing) suffer from serieous outlier problems (20% to 30% of the data are outliers). The results improve rapidly in the 2 and 3 hour processing, but it is an impending problem to identify the outliers. The results from the time period covering June 2000 show that we would have positively identified the size of the coseismic offset within a few hours. We see no indication of preseismic motions. IMO will continue to work on this.

Figure 1. Velocities for the CGPS stations in Iceland, assuming REYK is moving at velocity 9.6 mm/yr west and 2.1 mm/yr north (black arrows) compared to the NUVEL-1A plate motion model velocities (orange arrows). Confidence limits are at the 2 sigma level.

Figure 2. Time series of VOGS relative to REYK (black data points). Model fit shown in red. Displacements in east, north and up directions are defined as positive. Outliers have been removed. The offset in mid-2000 is due to the June 2000 earthquake sequence.