G43B-01 INVITED
Fusion of High-Rate GPS and Seismic Data: Applications to Early Warning Systems for Mitigation of Geological Hazards
We discuss the fusion of low-latency (1 s) high-rate (1 Hz or greater) CGPS displacements and traditional seismic data, in order to extend the frequency range and timeliness of surface displacement data already available at lower frequencies from space borne InSAR and (typically daily) CGPS coordinate time series. The goal is development of components of early warning systems for mitigation of geological hazards (direct seismic damage, tsunamis, landslides, volcanoes). The advantage of the GPS data is that it is a direct measurement of ground displacement. With seismic data, this type of measure has to be obtained by deconvolution of the instrument response and integration of the broadband (velocity) measurements, or a double integration of the strong motion (acceleration) measurements. Due to the bandwidth and the dynamic range limits of seismometers the accuracy of absolute displacements so derived is poor. This problem is not present in the high-sample rate GPS data. While the seismic measurement provides a powerful constraint on the much noisier GPS measurements, unlike the seismometer, the GPS receiver never clips. Using the Network for Earthquake Engineering Simulation (NEES) Large High-Performance Outdoor Shake Table at USCD, we present an example of combining in real-time 50 Hz GPS displacements and 250 Hz raw accelerometer data using a multi-rate Kalman filter, previously applied to bridge monitoring. A full-scale 7- story building atop the shake table was subjected to high intensity shaking by replaying the Sylmar accelerometer record from the Mw 6.7 1994 Northridge earthquake. The resulting 250 Hz displacement waveform is significantly more accurate than obtained solely by low-pass filtering and double integration of the 250 Hz accelerometer records. Next we demonstrate the elements of an earthquake early warning system by analyzing the 2003 Mw 8.3 Tokachi-Oki thrust earthquake off Hokkaido Island detected by the dense Japan national real-time CGPS network. The network has an approximately 20-km spacing with 1156 stations streaming 1 Hz data to a central facility. A Delaunay triangulation of the network is created every second and the 1 Hz displacements within triangular element are converted to principal components of strain to detect the event. The large spatial extent allows us to compute displacement waveforms relative to a station well away from the affected region. We then compute an earthquake source model using the displacement waveforms.
G43B-02 INVITED
The fifth model for the huge tsunami generation off northwest Sumatra during the 2004 Sumatra-Andaman earthquake
The 2004 Mw 9.2 Sumatra-Andaman earthquake caused a huge tsunami of more than 20 m on average along the west coast of Aceh. Four hypothetical models have been proposed for tsunami generation. The first model is that coseismic slip along the Sumatran megathrust is responsible for generation of the huge tsunami [e.g.,Henstock et al., 2006]. In this case, however, an additional tsunami generation mechanism such as inelastic deformation of soft accretionary sediment may be needed [Seno and Hirata, 2007]. The second model is that the most trenchward splay fault branching updip from the megathrust displaced coseismically [Soh et al., 2005; Seeber et al., 2007]. Splay faults can generate larger tsunamis, primarily because of their steeper dip. The third model is that the most landward splay fault, located at the eastern margin of the Sumatran outer-arc high, displaced coseismically [Sibuet et al., 2007]. The fourth model is that the West Andaman Fault, just west of the Aceh (forearc) basin, displaced coseismically [Plafker et al., 2005, 2006], though ROV diving surveys did not find any signature of coseismic fault motion along this fault [NT0502 scientific party, 2005; SEATOS scientific party, 2005]. Recent studies help constrain possible 2004 tsunami sources off northwest Sumatra. Coseismic slip limited to the deeper (landward) part of the megathrust cannot explain the observed Sea Surface Heights [Geist et al., 2007], indicating that the rupture reached the more trenchward portion of the fault. Calculated tsunami backward wave-fronts suggest that the trenchward boundary of the tsunami source off northwest Sumatra was located near the accretionary prism toe and that long-wave-approximated, maximum uplift area was located in the middle of the outer-arc high [e.g., Fine et al., 2005]. Array analysis of short-period tsunami dispersed waves, observed with hydrophone arrays in the Indian Ocean, suggests that the source is located at 4.3 degN and 93.8 degE [Hanson et al., 2007]. This location coincides with along Middle Thrust, depicted by Sibuet et al. [2007], which is located in the middle outer arc-high. The estimate is considered precise but it is possible that the source location may be a few tens of kilometers at most trenchward of this position. The source size was much smaller than 30 km in length[Hanson et al.,2007]. All the studies above provide critical constraints on the location and mechanism of anomalous tsunami generation offshore northwest Sumatra. Taking these constraints into account, we can construct a new model to explain the unusual tsunami generation off northwest Sumatra. We thus propose the fifth model that the 2004 earthquake ruptured updip along the megathrust (plate interface) near the deformation front, but branched onto one of the outer-arc high splay faults: either the Middle Thrust or possibly the Lower Thrust of Sibuet et al.[2007].
G43B-03 INVITED
Development of a new Tsunami Monitoring System Using a GPS Buoy
A tsunami monitoring system using a GPS buoy has been developed for more than ten years. Real-time kinematic (RTK) GPS technology was used for this purpose. After a series of preliminary experimental studies, the third experiment was conducted offshore Ofunato city, northern Tohoku, Japan. GPS antenna was set at the top of the buoy and the 1-sec sampling data were transmitted to the ground base of about 1.6km distance together with other ancillary data. The data was processed at the ground base and the estimated 3D positions were disseminated through internet. This system operated for about three years of 2001-2003 and succeeded to detect two tsunamis of about 10cm amplitude; 2001 Peru earthquake and 2003 Tokachi earthquake, by applying a simple filtering technique. After this successful experiment, the fourth system was newly designed and was established about 12km south of Muroto Promontory, southwestern Japan in early April 2004. The buoy has experienced nearby passages of several typhoons with a maximum wave of about 20meter and has shown a total integrity for an operational use. On September 5th 2004, a large earthquake of Mw7.4 occurred about 200km east of the buoy. The GPS buoy successfully recorded the tsunami with about 10cm amplitude at the first peak arrival of about 10 minutes before its arrival at the nearest coast of Muroto Promontory. The simulated record has shown excellent consistency with the observed tsunami, suggesting high potential for predicting tsunami height at the coast before its arrival, if the record is efficiently implemented in the tsunami warning system. The system has been adopted as a national sea-surface monitoring project and has been deployed at several locations around the Japanese coasts for monitoring also wind-waves.
G43B-04 INVITED
Tsunami Early Warning: Benefits from GNSS real-time data transmission
Nowadays GNSS technologies are used for a large variety of precise positioning applications. The accuracy can reach the mm level depending on the data analysis methods. GNSS technologies thus offer a high potential to support tsunami early warning systems, e.g., by detection of ground motions due to earthquakes. Although GPS-based precise ground motion monitoring is a standard method, it is not yet common to apply this technique under tight time constraints and, hence, in the absence of precise satellite orbits and clocks. The most challenging task in this context might be the direct detection of tsunami waves with GPS. A proper instrumental setup implies GPS-equipped buoy systems and tide gauge sensors. At the latter GPS can help to distinguish between real tsunami waves and apparent sea height changes due to co-seismic landmass movements. In our presentation we will show the sensor network and data communication layout chosen for the German Indonesian Tsunami Early Warning System (GITEWS). We will describe how data are collected, transferred and processed. This includes a view on the sensor system design, the communication system layout with real-time data streaming, the precise data processing strategy and the final products that are delivered to support the tsunami alert decision process.
G43B-05 INVITED
Recent Advancements in the U.S. Geological Survey Streamgaging Program
Accurate and timely information about streamflow is essential for the forecasting of floods and low flows, the
design and operation of water infrastructure, the delineation of flood plains, and the allocation and
management of water and environmental resources. Providing this information is no simple task. Streams
are highly dynamic systems undergoing nearly continuous variation in geometry and roughness and capable
of generating powerful hydrodynamic forces. These issues pose significant challenges.
The USGS streamgaging network addresses these challenges through a unique blend of sophisticated
sensing, recording, and telemetry equipment, and computer systems, coupled with a dedicated pool of field
hydrographers who visit the streams on a regular basis and during floods and droughts to make the essential
physical measurements and observations needed to convert automated in situ records into accurate
streamflow data. The hydrographers combine their observations with the telemetered data by processing,
computing, and applying adjustments to stage-discharge ratings, river-stage records, or both.
Although the basic streamgaging process was formulated in the 1890s and has changed only incrementally
for most of the intervening period, recent developments in hydroacoustic, RADAR, and LiDAR technology
and near real-time satellite telemetry have greatly improved the process, resulting in more accurate and
timely data. Emerging technologies, such as the use of noncontact microwave radar for measurement of
river stage and velocities promise significant advances in field monitoring. New approaches that utilize
correlative data other than river stage, such as in situ or surface velocity data, may further improve
streamflow records while reducing the need for on-site visits, and new software that automatically applies
some of the judgment now required of the field hydrographer, offers additional opportunities for improving the
data and controlling costs.
http://pubs.usgs.gov/fs/2007/3080/
G43B-06 INVITED
Real-time Geodetic Data for Earthquake and Volcano Hazards Applications: Current and Future Activities at the U.S. Geological Survey
The USGS Earthquake and Volcano Hazards Programs operate continuously-recording geodetic networks that include strainmeters, creepmeters, tiltmeters, and Global Positioning System (GPS) receivers. Data from some networks are available for real-time analysis while most networks provide measurements less frequently. Tiltmeter data are used to generate real-time volcanic deformation alerts. The full range of geodetic observations are used in estimating source parameters following earthquakes or volcanic events, although not in real-time. Methods for automatic detection of anomalous signals in daily GPS time series are being implemented. The increasing availability of real-time 1 Hz GPS observations suggests new applications for real-time data. For example, rapidly-available displacement estimates could improve the finite earthquake source models used by ShakeMap and provide input to a "slip sensor" for early warning of imminent seismic shaking. Real- time tracking of ongoing deformation measured by all of the above instruments and detection of anomalous changes would facilitate timely recognition and detailed monitoring of aseismic or volcanic events and the evolving hazard. Tools like these that fully utilize real-time geodetic data, if realized, would improve disaster- warning and response capabilities. Establishing reliable access to data in a form usable by analysis algorithms is currently one area of focus. The USGS is working to upgrade its existing GPS stations to real-time and is partnering with UNAVCO to create robust pathways for real-time high-rate GPS data from Plate Boundary Observatory sites, significantly expanding the potential coverage of monitoring and response tools. We are also exploring software solutions for processing high-rate GPS data in real-time. Diverse monitoring applications demand a wide range of features and capabilities. Software evaluation based on a variety of criteria including the ability to recover the amplitude and timing of known offsets with minimal outliers is ongoing. Continued research and development will focus on improving our understanding of source processes in order to refine analysis tools; assessing how real-time information will be monitored and used by scientists, emergency responders, and the public; and bringing tools that exploit these data to fruition.
G43B-07 INVITED
Data Latency Characteristics Observed Through Diverse Communication Links by the EarthScope USArray Transportable Array
The USArray Transportable Array has deployed over 600 stations in aggregate over the past four years. All stations communicate in near-real time using ip protocols over a variety of communication links including satellite, cell phone, and DSL. Several different communication providers have been used for each type of communication links. In addition, data are being acquired from several regional networks either directly from a data server or after passing through the IRIS DMC BUD system. We will present results about the latency of data arriving at the UCSD Array Network Facility where the real time data are acquired. Under normal operating conditions the median data latency is several seconds. We will also examine the data return rates through the near-real time systems. In addition we will examine the statistics of over 36,000 events which have automatic event locations and associations. We evaluate the timeliness of these results in the context of seismic early warning systems.
G43B-08
Real time data from the Plate Boundary Observatory continuous GPS network
EarthScope's Plate Boundary Observatory (PBO) runs a network of 1,100 continuous GPS stations in North America and has the potential to be a major provider of real-time GPS data for scientific research, hazard monitoring and survey control. PBO is planning to implement real time data flow for its three volcanic subnetworks (at Mt. Saint Helens and Alaksa's Akutan and Unimak Islands) to maximize the return of scientifically important data in the event of an eruption that destroys the installations. GPS sites with collocated instruments for meteorological measurement are also targeted for both GPS and met data streaming in the near future. On a larger scale, the USGS and a handful of academic institutions are doing research on integrating GPS into earthquake early warning (EEW) networks. The implementation of GPS- based EEW will involve real time streaming from GPS sites on major faults and in areas of high seismic hazard, and PBO is partnering with the USGS to help develop the first implementation of this early warning capability. Finally, planning is underway to develop open statewide real time networks to serve surveying communities and the general public, and PBO is positioned to be a key data provider for these efforts as well. PBO has been operating a pilot program to provide real-time GPS streams to the public from 75+ stations from the Salton Sea to Alaska. PBO's streaming data is provided exclusively via the NTrip protocol, from servers located at UNAVCO headquarters in Boulder, CO. The formats supported are BINEX and RTCM 2.3 at 1 second sampling, with RTCM 3.0 to be added in the near future. Access to PBO data streams is currently unrestricted and users are free to rebroadcast these streams provided they do not charge for these services. Our experience with this program indicates that we are technically capable of streaming real time GPS data from most of our network using existing telemetry, although PBO's IT infrastructure would have to be upgraded to support an expansion of the current system.