GGNNSSSS CCoonnttrriibbuuttiioonnss ttoo TTssuunnaammii EEaarrllyy WWaarrnniinngg SSyysstteemmss -- WWiitthh FFooccuuss oonn EEuurrooppeeBente Lilja Bye1, Halfdan Kierulf1, Tilo Schöne2, Oddgeir Kristiansen1

1 The Norwegian Mapping and Cadastre Authority, Norway (byeben@statkart.no), 2 GeoForschungsZentrum Potsdam, Germany

Tsunami Risk ANd Strategies For the European Region

To include GNSS in a warning system, especially tsunami early warning system (TEWS), require precise analysis in real time or close to real time, with robustness according to data quality and analysis strategies.

MethodesThere are two main approaches, using a global approach with global orbit (and clock) products or using a local vector approach.

With a local approach you need two or more GPS receivers in the area, where at least one is located outside the supposed deformation area of the earthquake. The advantages of this approach is: -you are independent of stations outside your network. -you may design a dedicated communication infrastructure for the actual area -you do not rely on other services, institutions or products. With a sufficient dense network it may give good and accurate results and as long as your reference station(s) is not affected by the accident itself you do not need global reference frames. However, a local vector solution will always depend on the reference station you use, especially during the earthquake you may be effected by deformation also on your reference station. Real or artificial distortion of the reference station will reflect in an apparent movement of the other station(s) and may be interpreted as an apparent incident.

For a warning system two different types of global approaches can be useful. The double difference (DD) or Precise Point Positioning (PPP). Both depend on availability of good global orbital products. For PPP you also need clocks.

The DD approach have many of the same feature as the local vector approach. However, you can use larger network and more remote stations reducing the effect of stations problems.

With the PPP approach you are independent of other stations which could be a large advantage for a warning service. However, your results strongly depend on existing orbit and clock products. The existence of such products are on of two main limiting factors of PPP.

Precise orbit and clock products is based on a global analysis of a network of GNSS stations. Analyzed by one or more analysis centers and then distributed to the users. This is a very demanding challenge for real time applications especially for the clocks. Orbits can be predicted with relatively good accuracy (for instance the IGS-ultra rapid orbits(IGU)), while clocks are largely unpredictable. EUREF has recently (see http://www.epncb.oma.be/_dataproducts/data_access/real_time/) established a service for real time clocks, to be used together with the IGU orbits. An alternative approach is to have a dedicated global analysis as a first step in the warning system analysis (GFZ is working with this for the Indian Ocean in the GITEWs project)

Analysis StrategiesAMBIGUITY The other limiting factors using PPP for warning purpose is related to the fixing of the phase ambiguity. Without fixed ambiguity the accuracy of GNSS processing results are limited when you only have short periods of data available. Traditionally, ambiguity resolution has been impossible in a PPP solution. Recently Ge et al. 2005 have developed a method to resolve ambiguity also for PPP solutions. The method is developed further by Geng et al. 2008. Also for local vector solutions and DD solutions the ambiguity fixing problem may by a limiting factor for early warning purpose. Phase break can very likely occur in relation to larger earthquake and it may be a problem to fix the ambiguity correctly within a limited time frame (see Figure 2).

EXAMPLES OF ANALYSIS STRATEGIES In Figure 1 and Figure 2, we compare different solutions from the Honshu 2008 earthquake. We have analyzed a network of 5 stations h912, h192, h193, h173 and h174. with two different softwares; GIPSY and GAMIT. With GIPSY a PPP solutions with the stations coordinates as stochastic variables. GAMIT is used with the track module with h912 as reference stations. Also the GIPSY solutions are relative to h912.

Time series for h193 relative h912. The results are from left GAMIT, GIPSY and the difference. Upper panels east and lower panels north component. GIPSY analysis by Rui M S Fernandez.

In Figure 1 we have plotted the station h193 which are very close to the earthquake and with very good data, from left GAMIT, GIPSY and the difference. We see a very good agreement between the two solutions before the earthquake and also the magnitude of the deformation agree. After the earthquake we see some minor differences in the solutions most likely due to different stochastic parameters.

In Figure 2 the results for stations h173 are plotted. The results are from left GIPSY, GAMIT and GAMIT with more data added. The station have a data gap after the earthquake, but no cycle slips. A standard solution with track assume that a gap of this magnitude imply a cycle slip. Using only about 10 minutes of data is not enough to resolve this ambiguity (mid panel of Figure 2).

Time series for h173 relative h912. The results are from left GIPSY, GAMIT and GAMIT with more data added. Upper panels east and lower panels north component. GIPSY analysis by Rui M S Fernandez.Conclusions

The ambiguity fixing problem as demonstrated in Figure 2 highlight one of the main issues for use of GNSS in TEWS. The question is: How do we make robust analysis strategies and statistical methods for unambiguity detection of crustal deformation in near real time? Problems related to outliers detection, cycle slips, large earth crust deformation, mechanical distortion of monument or technical problems, are all more likely to occur in relation to an earthquake. All of these factors need to be addressed to make reliable results that can be used in a TEWS. However, if these questions are solved, GNSS can give a valuable input to a TEWS. GNSS can precisely describe the geometry of the fault zones of the earthquake in near real time, which gives more precise and rapid determination of magnitudes as well as improved modeling of tsunami waves.

Geodesy is the science about Earth's geokinematics, gravity field and rotation and their changes with time. The changes in these 'three pillars' of geodesy are inseparably related to the dynamics of the Earth system. Studies of natural hazards, climate change, sea level variations and navigation are some examples of areas that not only benefit from but depend on geodetic observations.

The four elements of effective early warning systemsSource: The International Strategy for Disaster Reduction (ISDR)

Space-geodetic techniques and dedicated satellite missions are crucial tools in the determination and monitoring of geokinematics, Earth's rotation and gravity field. We need several techniques in order to determine one of the fundamental geodetic products, the terrestrial reference frame. Some of the techniques are listed below.

GNSSGlobal Navigation Satellite SystemMeasures positions and displacements of points. High temporal resolution.

SLRSatellite Laser RangingDetermines positions and Earth orientation parameters.

VLBIVery Long Baseline InterferometryDetermines positions, Earth rotation parameters and scale.

InSARInterferometric Synthetic Aperture RadarMeasure topography and topographic or surface changes. Low temporal resolution.

GOCEGravity field and Steady-State Ocean Circulation ExplorerDetermines the geoid, a height reference surface.

The main group of sensors we use in tsunami warning are seismic, marine and geodetic.

Earthquakes, landslides, volcanoes and precarious rocks/rockslides can all trigger tsunamis.

The three phases of a tsunami, pre-, co- and post event, requires different temporal and spatial monitoring and analysis.

References and acknowledgements:Ge, M., Gendt, G., Dick, G. and Zhang, F. P. (2005) Improving carrier-phase ambiguity resolution in global GPS network solutions,Journal of Geodesy, Volume 79, Issue 1-3, pp. 103-110

Geng, J.; Meng, X.; Teferle, F. N.; Dodson, A. H.; Shi, C.; Liu, J. (2008) Ambiguity Resolution in Precise Point Positioning for Sub-cm Precision With Hourly Data, American Geophysical Union, Fall Meeting 2008, abstract #G41C-0641

ISDR, Living with Risk - A global review of disaster reduction initiatives, 2004, International Strategy for Disaster Reduction, United Nations.

Plag, H.-P., 2006. National geodetic infrastructure: current status and future requirements - the example of Norway, Bulletin 112, Nevada Bureau of Mines and Geology, University of Nevada, Reno, 97 pages.

Plag, H.-P., 2006. GGOS and it user requirements, linkage and outreach, in Dynamic Planet - Monitoring and Understanding a Dynamic Planet with Geodeticand Oceanographic Tools, edited by P. Tregoning & C. Rizos, vol. 130 of International Association of Geodesy Symposia, pp. 711-718, Springer Verlag, Berlin.

Special thanks to Georg Weber, Loukis Agrotis, Carine Bruyninx and Frank Webb.

The International Association of Geodesy (IAG) has a number of services, one of them being the International Global Navigation Satellite System (IGS). Euref is a European sub network and service of IGS. A selection of Euref GNSS stations deliver data in real-time (see map below).

Analysis and graphics courtesy of J. Geng, University of Nottingham, UK. Left: Orbit comparison of ultra rapid (IGU) with final IGS orbits. Right: Clock comparisons of clocks with final IGS clocks.

Tests show that real-time data from Euref meet the TEWS requirements on temporal and spatial resolutions.

For sustainability, quality and economic reasons, a European Tsunami Early Warning System (or UN's NEAMTWS) geodetic sensors must be based on and expand through the existing geodetic network. Via the Global Geodetic Observing System GGOS, the European infrastructure will also be included in GEO's Global Earth Observing System of Systems.