G43A-0642
California Real Time Network: Test Bed for Mitigation of Geological and Atmospheric Hazards within a Modern Data Portal Environment
Global geological and atmospheric hazards such as earthquakes, volcanoes, tsunamis, landslides, storms
and floods continue to wreak havoc on the lives of millions of people worldwide. High precision geodetic
observations of surface displacements and atmospheric water vapor are indispensable tools in studying
natural hazards along side more traditional seismic and atmospheric measurements. The rapid proliferation
of dense in situ GPS networks for crustal deformation studies such as the Earthscope Plate Boundary
Observatory provides us with unique data sets. However, the full information content and timeliness of these
observations have not been fully developed, in particular at higher frequencies than traditional daily
continuous GPS position time series. Nor have scientists taken full advantage of the complementary nature
of space-based and in situ observations in forecasting, assessing and mitigating natural hazards. The
primary operating mode for in situ GPS networks has been daily download of GPS data sampled at a 15-30 s
sample rate, and the production of daily position time series or hourly tropospheric zenith delay estimates.
However, as continuous GPS networks are being upgraded to provide even higher-frequency information
approaching the sampling rates (1-50 Hz) of modern GPS receivers, and with a latency of less than 1
second, new data processing approaches are being developed. Low-latency high-rate measurements are
being applied to earthquake source modeling, early warning of natural hazards (geological and atmospheric),
and structural monitoring. Since 2002, more than 80 CGPS stations in southern California have been
upgraded to a 1 Hz sample rate, including stations from the SCIGN and PBO networks, and several large
earthquakes have been recorded. The upgraded stations comprise the California Real Time Network (CRTN
- http://sopac.ucsd.edu/projects/realtime/). This prototype network provides continuous 1 Hz (upgradable to
10 Hz at some stations) GPS relative displacements and troposphere delay estimates with a latency of less
than 1 s. Steps are being taken to tie these higher-order data products to the global (ITRF) reference frame,
and discussions are underway to extend this capability throughout California by upgrading more
UNAVCO/PBO stations. With funding from NASA, CRTN provides a test bed for developing advanced in situ-
based observation systems within a modern data portal environment, which can be extended seamlessly to
the entire PBO region and to other plate boundaries. I describe a prototype early warning system for
earthquakes using CRTN, which is also being deployed at other plate boundaries. I show how researchers
can access advanced observations of displacements and troposphere delay in real-time and replay
significant events within the GPS Explorer data portal and Geophysical Resource Web Services (GRWS)
environment.
http://sopac.ucsd.edu/projects/realtime/
G43A-0643
Mapping Hurricane Inland-Storm Tides
Historically, hurricane-induced storm-tides were documented through analysis of structural or vegetative
damage and high-water marks. However, these sources rarely provided quantitative information about the
timing of the flooding, the sequencing of multiple paths by which the storm-surge waters arrived, or the
magnitude of waves and wave run-up comprising floodwaters.
In response to these deficiencies, the U.S. Geological Survey (USGS) developed and deployed an
experimental mobile storm-surge network to provide detailed time-series data for selected hurricane landfalls.
The USGS first deployed the network in September 2005 as Hurricane Rita approached the Texas and
Louisiana coasts. The network for Rita consisted of 32 water-level and 14 barometric-pressure monitoring
sites. Sensors were located at distances ranging from a few hundred feet to approximately 30 miles inland
and sampled 4,000 square miles. Deployments have also occurred for Hurricanes Wilma, Gustav, and Ike.
For Hurricane Gustav, more than 100 water level sensors were deployed.
Analysis of the water-level data enable construction of maps depicting surge topography through time and
space, essentially rendering elements of a 3-dimensional view of the storm-surge dome as it moves on-
shore, as well as a map of maximum water-level elevations. The USGS also acquired LIDAR topographic data
from coasts impacted by hurricanes. These data reveal extreme changes to the beaches and barrier islands
that arise from hurricane storm surge and waves. By better understanding where extreme changes occur
along our coasts, we will be able to position coastal structures away from hazards.
http://water.usgs.gov/osw/programs/storm_surge.html
G43A-0644
Status of the Canadian Pilot Project for a GPS-Augmented Tsunami Warning System
Whether or not a tsunami has been generated from a large near-shore earthquake cannot be determined within a crucial time window from seismic data alone. Crustal displacements observed for past subduction- thrust earthquakes show that coseismic motions of the Earth's surface, even hundreds of kilometres from the fault, can be used to determine the nature of the rupture and whether the earthquake is capable of generating a large tsunami. By adapting existing continuous GPS stations of the Western Canada Deformation Array (WCDA) to stream 1Hz data in real time, the Geological Survey of Canada has established a small prototype network to facilitate real-time positioning along the coast of the Canadian segment of the Cascadia subduction zone. This prototype network has now been operational for three years. The network uses geodetic-quality GPS installations, a mix of land-based and satellite-based IP connections for data streaming, and the RTD real-time positioning software. Current precisions in relative site positions are 2-3 cm horizontally and 7-10 cm vertically for independent 1-sec solution epochs. These precisions are limited primarily by the length of baselines, which all exceed 50 km, but are clearly adequate to resolve differential motions (~ 2m horizontal; ~1m vertical) expected for a megathrust earthquake immediately offshore. Other issues which have proven critical include the quality of satellite tracking and the robustness of the communication links. We have developed a basic web-browser interface which allows real-time remote monitoring of changes in position of the network sites. The objective is to facilitate immediate access by the Tsunami Warning Centres to the coseismic site displacements (magnitudes and patterns) and, in conjunction with seismic data, allow them to rapidly determine the likelihood of tsunami generation.
G43A-0645
Testing a Collocated GPS and Tide Gauge Station at Mayagüez, Puerto Rico for Improved Tsunami Monitoring
The Puerto Rico Seismic Network (PRSN) of the University of Puerto Rico at Mayagüez is responsible for providing the Commonwealth of Puerto Rico and the Virgin Islands with both earthquake and tsunami information and warnings. Destructive tsunamis, as in the past, could be generated by large local earthquakes associated with the interaction of the North American and Caribbean plates. As part of the ongoing activities to improve both monitoring and detection capabilities for tsunami events, real time high- rate GPS and tide gauge networks have been established to complement the traditional seismic surveillance activities of the PRSN. One of the new GPS stations has been collocated with a tide gauge station in Mayagüez, Puerto Rico. In addition to providing improved geodetic information for the tide gauge, we will also test its applications for tsunami warning. A comparison of the time series of the GPS and tide gauge sensors will be presented and the feasibility of integrating these two data sets in near real time for warning purposes will also be discussed.
G43A-0646
A tsunami early warning system and tsunami source propagation
The present-day's tsunami early warning systems (TEWSs) assume that a tsunami source is created instantaneously over the whole source area. However, finite rupture propagation speeds reasonably affect tsunami generation and propagation. Travel times of the first tsunami waves generated from rupture propagation fault models have to arrive later at some regions than those from instantaneous fault models but do not arrive faster everywhere [e.g., Neetsu et al., 2005]. On the other hand, the effect of earthquake rupture propagation on tsunami amplitude has scarcely investigated based on real bathymetry, although it had studied under flat ocean assumption [Aida, 1969; Yamashita and Sato, 1974]. We investigated it for regional tsunamis (travel distance between 5deg and 30 deg) by using real bathymetry in terms of numerical simulation. The grid size is 3 minutes of arc, interpolated from ETOPO2 [Smith and Sandwell, 1997]. A constant amount of slip was given on each subfault. Other subfault parameters are the same as Hirata et al.[2006]. Effect on the first upward wave amplitudes of offshore tsunamis is summarized as follows; (1) The slower the rupture velocity, the larger the increasing rate of the tsunami wave amplitude. For instance, in the offshore of the east coast of Sri Lanka, the 1st tsunami amplitude increases 10% at the rupture velocity of 3.5 km/sec, 14 % at 2.5 km/sec, 23% at 1.5 km/sec, and 38% at 1.0 km/sec in the case of 500km-long fault. This indicates that the rupture propagation effect on regional tsunami amplitude is large in the cases of slow earthquakes. (2) Increasing (or decreasing) rate in tsunami amplitude due to the rupture propagation is not necessarily larger in longer fault than shorter fault. In other words, even in the case of a short fault, say a few hundred kilometers, the effect of the rupture propagation on tsunami amplitude may not be neglected. This is because the increasing (or decreasing) rate in tsunami amplitude is related to whether the arrival timings of subfault tsunami components are in-phase or out-of-phase, which is strongly depending on not only rupture velocity but also combination of tsunami travel time from a set of subfaults to the station. Finally, I would like to discuss whether the effect of the earthquake rupture propagation should be included in the next generation of the tsunami forecast/warning system. (A) Because normal earthquake ruptures propagate at 2 to 4 km/sec, the assumption that tsunami is generated instantaneously over the whole tsunami source would not be realistic. However, a method for estimating rupture propagation velocity on real- time base has not yet developed. Therefore any next tsunami forecast/warning system(s) may assume a priori given, constant rupture velocity (say 3 km/sec or/and 2 km/sec), instead of infinite rupture velocity, until real-time estimation of rupture propagation velocity will be devised. Such next system may consider both of unilateral and bilateral rupture propagations. (B) Coastal shallow bathymetry strongly affects coastal tsunami amplitude, perhaps than the effect of the earthquake rupture propagation; coastal shallow bathymetry can amplify tsunami amplitude a few times easily but the effect of the earthquake rupture propagation can do it only at most 50% or so. Only if high accurate, coastal shallow bathymetry has already been obtained, therefore, the use of information on actual rupture velocity, inferred from any real-time seismic/offshore tsunami inversion technique, would be effective. Otherwise an assumption of the a priori given, constant rupture velocity seems to be enough.
G43A-0647
Tsunami Genesis Theory and GPS Detection Methodology
Conventional tsunami warning systems often rely on estimates of an earthquake's magnitude to determine
whether a large tsunami will be generated. However, earthquake magnitude is not always a reliable indicator
of tsunami potential. This study includes two parts: 1) understanding how ocean receives energy from an
undersea earthquake during the tsunami excitation period to generate tsunamis and 2) developing a real-
time methodology to estimate the energy by using data from coastal GPS stations near the epicenter before
the tsunami reaches coastal areas. Specifically, the tsunami energy, derived from GPS displacements near
the epicenter, is used to calculate a new measurement called a "tsunami scale" ranging from one to 10, much
like the earthquake "Richter Scale", which is then used to discriminate earthquakes capable of generating
destructive tsunamis from those unlikely to. This method has been successfully tested on several historical
events of earthquake-generated tsunamis.
http://science.jpl.nasa.gov
G43A-0648
Rapid assessment of Cascadia tsunamis from real-time PANGA GPS crustal deformation measurements
Cascadia's natural hazards include earthquakes, tsunamis, volcanic eruptions, landslides, and tectonic subsidence along its coasts and inland waterways exacerbated by sea-level rise. The Pacific Northwest Geodetic Array, now comprised of nearly 200 continuous GPS receivers, has been deployed over the last two decades to focus exclusively on mitigating these hazards. In addition, over 150 receivers of the EarthScope Plate Boundary Observatory have also been installed in Cascadia, thus comprising a combined network of over 350 instruments. Of the 200 PANGA stations, nearly 140 are high-rate, real-time telemetered receivers mounted on CWU-built, tectonics-grade monuments. These stations straddle active crustal faults, volcanoes and landslides, they span the megathrust forearc and tsunamigenic regions along the Pacific coast, and they monitor ageing man-made structures such as dams, levees and elevated freeways. All data from this array, currently at over 140 stations, is streamed in real-time into CWU where it is archived and processed with JPL's GIPSY software. In 2005 PANGA received support from NASA, NSF and the USGS to implement real-time processing in support of mitigating Cascadia's natural hazards. We have implemented Trimble Navigation's proprietary RTK software and network monitoring software on all 140 stations, and specific parameter estimation routines on a subset of these stations. Pending available funding, we are also working to implement processing of this data with the RTGipsy software, which produces position time series within a global, not local, reference frame. We are currently writing applications that will facilitate rapid recovery during and after a large seismic event, tsunami, or volcanic eruption. These applications are focused on: • Inverting GPS deformation measurements for earthquake fault location, size, and slip distribution; • Using slip distributions to predict tsunami magnitude and run-up estimates; • Real-time monitoring of volcanic unrest and landslide early warning, as well as critical infrastructure such as dams and bridges; • GPS-based rapid estimation from afar (central Washington) of strong ground motion from large Puget Sound or megathrust earthquakes that incapacitate local seismic networks. These measurements will then be transmitted in real-time to the seismic networks and state emergency managers to facilitate emergency response; • Training of undergraduate and graduate students in high-precision geodetic measurement techniques and analyses needed for future hazards assessment and mitigation. Although the construction phase of the real-time network is nearing completion, the data analysis aspect is still ongoing. An overview of the network, data analysis strategies, parameter estimation routines, and third- party data access options will be discussed.
G43A-0649
Introduction to the High-Rate GPS Network of Puerto Rico and the U.S. Virgin Islands
The Puerto Rico Seismic Network at the University of Puerto Rico at Mayagüez is a regional earthquake and tsunami monitoring institute. One of its primary objective is to provide timely and reliable earthquake and tsunami information and warning to the state (Puerto Rico) and local governments, the US and British Virgin Islands, as well as to the general public. In the past five years, it has been expanding its operations for the establishment of a Caribbean Tsunami Warning Center. With funding of the Puerto Rico government and NOAA, it is operated 24 hours per day and 7 days per week. Broadband seismometers are generally unable to capture the full bandwidth of long period ground motions following very large earthquakes. As a result, it is difficult to rapidly estimate the true magnitudes of large earthquakes using only seismic data. High-rate GPS has been justified as a very useful tool in recording long-period and permanent earthquake ground motions. Estimation of the true magnitude (and therefore tsunami potential) of large earthquakes may be determined more accurately in a timely manner (minutes after the quake) using high rate GPS observations. With the major aim of improving the ability of the PRSN in rapidly and precisely monitoring large earthquakes, NSF funded a Major Research Instrumentation (MRI) project, Acquisition of 9 High-rate GPS Units for Developing a Broadband Earthquake Observation System in Puerto Rico and the U.S. Virgin Islands (EAR-0722540, August 1, 2007—July 31, 2009). The major purpose of this project is to build a high-rate GPS network in Puerto Rico and the U.S. Virgin Islands. The GPS network includes 3 campaign and 6 permanent GPS stations. These campaign stations were designed to use in emergency response after large earthquakes to get co-seismic and post-seismic displacement. These six permanent stations were designed to complement current seismic observation system of Puerto Rico and U.S. Virgin Islands. We have installed three permanent GPS stations in May, 2008. They locate in Arecibo Observatory, Bayamon Science Park, and Caja de Muertos Island. We will install the other three stations in October, 2008. They will be located in Mona, Culebra, and St. Thomas islands. All of these permanent GPS stations are colocated with seismic stations operated by the Puerto Rico Seismic Network and the Puerto Rico Strong Motion Program. They are also very-closely spaced to the Tide Gauge stations operated by PRSN and NOAA. Therefore they will also complement the tide gauge sea-level observation system to get accurate absolute sea-level changes after large earthquakes. The integrated velocitymeter-accelerometer- GPS earthquake observation system will advance knowledge of seismic wave propagation, the kinematics and dynamics of fault rupture process, pre-seismic, co-seismic and post-seismic deformation, and is also likely to be useful for improving building and critical structure designs. It will support earthquake and tsunami hazards research and mitigation in Puerto Rico and the surrounding region. High-rate GPS observations can also be used for real time tropospheric water vapor tomography which is useful for weather prediction, including improved hurricane track forecasting. Raw GPS data are freely available through the UNAVCO archive. As a result, a large number of researchers can potentially benefit from the data for research and applications ranging from neotectonics to atmospheric science to civil engineering.
G43A-0650
Capturing crustal deformation signals with a new high-rate continuous GPS network in Iceland
Iceland straddles the plate boundary of the North American and Eurasian plates. Several volcanic rift zones and two main fault zones accommodate the plate spreading across the island. Monitoring the high level of activity and understanding the complex tectonics requires a dense network of geodetic stations. The continuous GPS (CGPS) network in Iceland is rapidly expanding. From 1999 to 2005 the network grew from two IGS stations to about 20 stations, operated by the Icelandic Meteorological Office (IMO). Our project was initiated in 2006 with the aims to improve the temporal and spatial resolution of the plate spreading in Iceland, as well as increase our understanding of volcanic and tectonic activity. As of September 2008, 34 new sites with 1s sampling rate are operating and some have been collecting data since 2006. The network is densest in the seismically active areas in the South Iceland Seismic Zone (SISZ), the Reykjanes Peninsula and across the Tjörnes Fracture Zone (TFZ) in North Iceland. The network in northern Iceland will help constrain the partitioning of motion on the active seismic lineaments that comprise the TFZ and their interaction with the Northern Volcanic Zone. Many of the new sites are located in central Iceland in a region that is suggested to be a micro-plate, termed the Hreppar block. Data from the new stations will help us determine whether the area is a stable block or should be classified as a plate boundary zone. A dense network of stations has been installed at Hekla volcano, as well as new sites near other recently active volcanoes in Iceland: Grimsvötn, Bárdarbunga and Katla. In May 2008 a M6.3 earthquake occurred in the western part of the SISZ. We observed about 15 cm of horizontal motion at the two closest CGPS stations, located on either side of the activity. A study of the May 2008 event will be presented in Session T29.
G43A-0651
Love Wave Dispersion in Central North America Determined Using High Rate GPS Absolute Displacement Seismograms
The increasing number and reliability of high-rate GPS networks offer the opportunity to augment exiting seismic network coverage by using GPS time series to measure large amplitude, weak motion ground displacements. HRGPS time series are orders magnitude more noisy than seismic and are also differential rather than absolute in nature. We focus on displacements from the Mw 9+ Great Sumatra-Andaman earthquake of December 26, 2004, as recorded by 94 randomly spaced HRGPS stations located mostly in central North America. We demonstrate that stacking of HRGPS time series can be used to recover absolute GPS ground displacements. We then use beamforming without spatial or sidereal filtering to recover the fundamental continental Love phase velocity dispersion curve for central North America. Results agree well with those obtained from over 30 broadband seismic recordings between 20 and 300 second periods, at which point the broadband seismic response rolls off. The HRGPS data continue to provide useful results until about 500 seconds, beyond which wavelength exceeds the array aperture and other analysis methods, such as wave gradiometry, might be used. Our results suggest that filtering of HRGPS time series in the frequency and spatial domains will provide a powerful GPS multipath reduction technique not dependent upon repeatability of GPS multipath or constancy of GPS station site conditions.
G43A-0652
High rate GPS data on active Volcanoes: The Mt. Etna real time GPS network- ETN@NET
The INGV (Istituto Nazionale di Geofisica e Vulcanologia) branch of Catania (Italy) manages a network of 24 CGPS stations on Mt. Etna for volcano monitoring purpose. The aim of such network is to reduce the hazard related to the presence of human settlements potentially exposed to lava flows from the central craters or from lateral intrusions. Since 2004, 18 of these stations transmit data in real time to the master station of Catania. Collected data are processed by using the Geodetics's RTD software. In this work we present the software tool we have developed for managing and processing high rate GPS time series and some examples of data related to volcanic activity. Moreover we have attempted innovative approaches in order to thoroughly investigate the high rate GPS signal. In particular we have developed some novel filtering techniques able to remove typical noise sources (of various origins) from the signal. The synthesis of our work is aimed 1) to a characterization of high rate GPS time series on active volcanoes in both frequency and time domains and 2) to design real time risk assessment applications based on our findings, joint with other geophysical signals (e.g. broadband seismic signals).
G43A-0653
Tests of RTG (Real Time GIPSY) for Earthquake Early Warning and Response Applications in Southern California
Recent developments in high-rate real-time GPS technology and processing promise to improve the application of GPS to earthquake early warning and response. Point positioning processing algorithms, which do not require a reference station, are particularly attractive for these applications since any reference station will itself be displaced during a large earthquake. USGS Pasadena is testing one such software package, Real Time GIPSY (RTG), developed and supported by the Jet Propulsion Laboratory (JPL). JPL uses RTG for precise real-time satellite orbit and clock determination, formats the results as corrections to the GPS broadcast orbit, and provides a real-time stream over the Internet. In our tests we use a locally- installed copy of RTG to compute real-time positions of GPS stations at a sampling rate of 1 second. In clean sections of the position time series are good, with rms scatter of 2 to 4 cm in the north and east components, and 5 to 10 cm in the vertical. Current work is designed to understand and handle occasional convergence delays and large outliers; many outliers repeat every sidereal day and may be correlated with multipath or with the rising or setting of individual satellites. The test site is in a less-than-ideal setting, and we are experimenting with the software setup and with different sites with fewer sources of multipath and better sky view. USGS Pasadena currently operates about 90 permanent continuously-operating GPS stations, about 20 of which are real-time. With funding from the USGS MultiHazards Demonstration Project, USGS Pasadena is cooperating with the California Integrated Seismic Network to co-locate approximately eight real-time GPS receivers at new seismic stations along the southern San Andreas fault. The Plate Boundary Observatory (PBO) is also converting many of its southern California stations to real-time operation. These real-time data and software such as RTG promise to improve USGS Pasasdena's geodetic response to large southern California earthquakes.
G43A-0654
Potential and Pitfalls of High-Rate GPS
With completion of the Plate Boundary Observatory (PBO), we are poised to capture a dense sampling of
strong motion displacement time series from significant earthquakes in western North America with High-Rate
GPS (HRGPS) data collected at 1 and 5 Hz. These data will provide displacement time series at potentially
zero epicentral distance that, if valid, have great potential to contribute to understanding earthquake rupture
processes. The caveat relates to whether or not the data are aliased: is the sampling rate fast enough to
accurately capture the displacement's temporal history? Using strong motion
recordings in the immediate epicentral area of several 6.7
G43A-0655
Analysis of High Rate GPS Data From the 2007 Mw 7.7 Tocopilla Earthquake in Northern Chile
The CANTO (Central ANdean Tectonic Observatory) geodetic network captured the November 15 Tocopilla earthquake. Using GAMIT/TRACK software we compute 3-component 5Hz time series of station displacement to investigate coseismic and postseismic kinematics. We present spectral characteristics of the recorded coseismic waveforms and compare them to those of other earthquakes (southern Sumatra events). We determine arrival times and estimate simple source parameters from the high resolution records.
G43A-0656
High rate GPS solutions from 2007 Mw7.7 Tocopilla, Chile Earthquake
The Mw 7.7 Tocopilla, Chile earthquake occurred within the CANTO (Central ANdean Tectonic Observatory) geodetic network, with ten stations recording at 5Hz. Because of the magnitude and proximity of the event, many of the stations recorded the dynamic as well as the static displacement from the earthquake. In order to provide the best time series for source modeling studies (see Minson et al, Fall AGU 2008), we have tested several high-rate analysis strategies. The high-rate solutions were analyzed in a baseline mode, fixing the position and clocks of a remote reference site, and estimating satellite clocks, tropospheric parameters, as well as the station position and clock of the non-stationary site. The station position was estimated at longer time intervals leading up to the onset of earthquake displacement (e.g., 10 minutes) to allow for a more robust position and troposphere estimate. During the earthquake, station positions were estimated at 5Hz. The station position estimates were reduced to 1 second and then 1 minute with time following the earthquake. We mitigated the effects of multipath by applying a modified sidereal filter correction to the phase observables. The multipath correction was estimated from static solutions based on 15 second data from the days prior to the earthquake. Given the potential for afterslip and aftershocks, we avoided using data from after the earthquake for estimation of the multipath correction. The multipath correction significantly reduced noise in the time series at the several minute to several hour period. We also compare 1Hz, and 5Hz solutions from the CANTO stations. While for some stations these solutions are very similar, there are other stations where the 5Hz solutions captures higher rate motion that were aliased in the 1Hz solutions. This result is a strong argument for increasing GPS data collection rates when looking for dynamic seismic motion.
G43A-0657
Source Process of the 2007 Mw 7.7 Tocopilla, Chile Earthquake
The magnitude 7.7 Tocopilla earthquake of November 14, 2007 ruptured a stretch of the northern Chile subduction zone just north of the 1995 Antofagasta earthquake source region. In this study, we compare a Bayesian analysis with a simulated annealing inversion methodology to determine a finite fault rupture model for the Tocopilla earthquake. While we also incorporate both teleseismic data and satellite radar images, high-rate (5 samples/sec) GPS data provides particularly important constraints on the source process of this event. Because the Tocopilla earthquake was located immediately coastward of the CANTO (Central ANdean Tectonic Observatory) geodetic network, the evolution of the local slip field is well-recorded. We find that the high-rate continuous GPS data is key to determining the location of the hypocenter and asperities as well as the evolution of slip. Most of the slip is located east of the coastline with at least two distinct regions of concentrated slip.
G43A-0658
Far-field signals on 2008 Mw=8.0 Wenchuan Earthquake from GPS seismometer and TEC
The 2008 Mw=7.9 Wenchuan earthquake occurred at the eastern edge of the Tibetan plateau on May 12, killing thousands of people in several cities along the western Sichuan basin in China. After such a large- magnitude earthquake, rearrangement of stresses in the crust results in subsequent damaging earthquakes Mw=1-6 along the boundary between the Longmen Shan and Sichuan basin. However, real time observation data are quite sparse near this earthquake. In this paper, far-field signals on the main shock of this earthquake are extracted and investigated from national continuous GPS network observations and ionospheric delays. GPS co-seismic total electron content (TEC) disturbances show a surface Rayleigh waves propagation at mean velocity of about 2.5 km/sec. In addition, high-rate (1-30 s) GPS data from 15 stations throughout China have been analyzed to derive displacement waveforms for this event. The average precision of 0.8 cm in the horizontal and 1.5 cm in the vertical are obtained after noise filtering, and large dynamic displacements are observable by GPS. Next step, we will further analyze GPS seismology recordings by comparing with broadband seismic recordings (integrated to displacement) as well as dynamics.
G43A-0659
Wave Propagation in the Ionosphere Associated With Earthquakes Revealed by GPS- TEC 4D Tomography
Hi-density high-rate sampling GPS network data is ideal for imaging quickly changing 3D structures in the ionosphere. GPS-TEC observation by GEONET in Japan during the 2003 Tokachi-Oki earthquake shows a clear propagating ionospheric disturbance. Heki and Ping (2005) interpreted the phenomena as a propagating sound wave in the thermosphere which was originally radiated into the atmosphere from the earthquake source region. To understand the ionosphere disturbance directly, we developed 4D ionosphere tomography method as an extension of a mantle tomography method to retrieve 3D seismic velocity structure of the mantle from traveltimes of seismic body waves from earthquakes to the seismic stations. We applied this tomography method to 1 Hz GPS-TEC data from GEONET which provides a dense line of sight coverage of space and time above and around the Japanese islands during and the after the earthquake. The image results show dispersive propagating waves, i.e., the phase speed of the waves is different from the wave energy propagation speed. The first phase, which appears first 100 km above the epicenter area, propagates horizontally with a phase speed about 1km/s and the secondary phase propagates slower. A close examination of the propagation of the first phase shows dispersion of the phase. The positive peak of the first phase travels 10% faster than the negative peak so that the peak shape broadens as it proceeds. The amplitude of the positive peak diminishes as it propagates over 1400 km distance from the source region. In contrast to the positive peak, the negative peak first appears as small amplitude and grows after traveling over 1000 km from the source region. Study of the evolution of the 4D GPS-TEC disturbance will provide rich information about the mechanisms of generation and propagation of ionospheric disturbance through the solid-earth-atmosphere-ionosphere coupling. Ionospheric disturbance can be generated from land surface deformation and the ocean surface deformation associated with shallow large earthquakes. An ultimate goal is to retrieve the earthquake fault parameters by modeling the GPS-TEC observations.
G43A-0660
Modelling the GPS-TEC observations after large earthquakes
Ionospheric perturbations after large and shallow earthquakes are now commonly monitored by GPS with a current mimimum magnitude of 6.5. In theory, the Total Electronic Content maps detected by a dense GPS network should present a similar pattern : at near field, the propagation of a slow wave (~1km/s) is related to the sound pulse generated at the epicentre, while at far field the fast (~3.5 km/s) wave train is excited by the Rayleigh surface waves. However, the satellites geometry is a key point for the observation of the two kinds, taking into account the 3D structure of the waves and the integration on the satellite-receiver ray-path. Then, we present the simulation results of the integrated electronic density pertubed by an acoustic plume for various ionospheric and geomagnetic conditions. The simulations are finally compared to the ionospheric perturbations monitored by the japanese GPS network GEONET just after recent major events : the Niigata-Chuetsu Oki earthquake of July 16, 2007 and the Iwate-Miyagi inland earthquake 2008 will be analysed here.
G43A-0661
Completion of the Southern California Plate Boundary Observatory GPS Network and Implementation of the Low Latency Salton Trough Network
Between 2003-2008 875 permanent PBO GPS stations have been built throughout the United States. Concomitant with construction of the PBO the majority of pre-existing GPS stations that meet stability specifications have been upgraded with Trimble NetRS and IP based communications to PBO standards under the EarthScope PBO Nucleus project. In October 2008, with completed construction of the Plate Boundary Observatory, more than 1100 GPS stations now share common design specifications and have identical receivers with common communications making it the most homogeneous geodetic network in the World. Of the 875 total Plate Boundary Observatory GPS stations, 216 sites are distributed throughout the Southern California region. 102 of the sites are built as SDBM, 111 DDBM, and 3 as strainmeter GPS hybrids. Fifteen second data is archived for each station and 1 Hz and 5 Hz data is buffered to be triggered for download in the event of an earthquake. Additionally, 125 of the existing former-SCIGN GPS stations have been integrated into the SoCal region of PBO, of which 25 have real-time data streams. The Salton Trough Radio Network (STRN) comprises of 20 stations equipped with Ethernet bridge Intuicom EB6+ (900 MHz) radios to transmit a high rate low latency data stream from each permanent GPS site. The high-rate low latency UStream data will be available to researchers who are developing prototype earthquake early warning systems in Southern California. A goal of the STRN is to make the data available rapidly enough for GPS-derived coseismic and dynamic displacements to be integrated into early warning system earthquake models. The improved earthquake models will better assist emergency response. UStream data will also aid surveyors who wish to use PBO GPS stations as permanent, high-quality base stations in real- time kinematic surveys. Requests for streaming data access at available GPS sites and latency statistics are located at http://pboweb.unavco.org/?pageid=107. Collectively, the cross fault spatial distribution of these 341 GPS stations in the seismically active southern California region has the grand potential of augmenting a strong motion earthquake early warning system. The installation and telemetry of 216 GPS site and upgrades of an additional 125 stations in Southern California was achieved by a major effort by field engineers, permitting staff, data personnel, drilling contractors and oversight by PBO management. It was a job well done.
G43A-0662
Change and Anomaly Detection in Real-Time GPS Data
The California Real-Time Network (CRTN) is currently generating real-time GPS position data at a rate of 1-2Hz at over 80 locations. The CRTN data presents the possibility of studying dynamical solid earth processes in a way that complements existing seismic networks. To realize this possibility we have developed a prototype system for detecting changes and anomalies in the real-time data. Through this system, we can can correlate changes in multiple stations in order to detect signals with geographical extent. Our approach involves developing a statistical model for each GPS station in the network, and then using those models to segment the time series into a number of discrete states described by the model. We use a hidden Markov model (HMM) to describe the behavior of each station; fitting the model to the data requires neither labeled training examples nor a priori information about the system. As such, HMMs are well suited to this problem domain, in which the data remains largely uncharacterized. There are two main components to our approach. The first is the model fitting algorithm, regularized deterministic annealing expectation- maximization (RDAEM), which provides robust, high-quality results. The second is a web service infrastructure that connects the data to the statistical modeling analysis and allows us to easily present the results of that analysis through a web portal interface. This web service approach facilitates the automatic updating of station models to keep pace with dynamical changes in the data. Our web portal interface is critical to the process of interpreting the data. A Google Maps interface allows users to visually interpret state changes not only on individual stations but across the entire network. Users can drill down from the map interface to inspect detailed results for individual stations, download the time series data, and inspect fitted models. Alternatively, users can use the web portal look at the evolution of changes on the network by moving backwards and forwards in time.
G43A-0663
BINEX as a Format for Near-Real Time GNSS and Other Data Streams
BINEX, for "BINary Exchange", is an open and operational binary format for GNSS data. It has been available
as a format option on several different GPS receivers from several manufacturers starting with Ashtech's
microZ in 2000, and has evolved to support GNSS data on Trimble's receivers. The data structures are very
compact and are organized in epoch-by-epoch records which do not rely on any prior records for decoding.
Typically, only a few hundred bytes per epoch are needed to store the L1 and L2 phase and code
pseudoranges (both to 1mm resolution), CNo measurements (to 0.1 dBHz resolution), loss-of-lock flags, and
so on. Ancillary site data, such as meteorological observations, can also be stored as BINEX records.
Each BINEX record also identifies whether it is of little-endian or big-endian construction, so that BINEX
creation can be optimized by processor type in a GNSS receiver or later construction by computer. Each
BINEX record also has a scaled checksum or CRC of 1-16 bytes, dependent on record length.
The Plate Boundary Observatory is currently using near-real time BINEX streams from Trimble NetRS
receivers as a means of outputting various ancillary site data. For example, meteorologic data, pore
pressure, borehole tilt, and so on can be monitored by multiple serial I/O on the NetRS and these port
queries bundled as BINEX records are directed to one or more BINEX output streams, in addition to the
primary GPS data epochs. Users can tap into which ever stream meets their need. In addition, the BINEX
records are stored in the NetRS in session files for later retrieval in case of real-time data loss in the
transmitted streams.
http://binex.unavco.org
G43A-0664
Early Source Characterization of Earthquakes Using the Back Projection Method
Rapid and accurate assessment of source (rupture) characteristics of a moderate to large earthquake can
be used for hazard assessment and guidance of emergency services, which are critical for efficient rescue
efforts to mitigate economic damages and/or loss of human lives, in addition to providing valuable inputs to
scientific community regarding details of source propagation. Earthquake source processes are typically
determined by inversion, which requires significant computational power and time. However, with the recently
developed Back Projection Method (BPM) (e.g. Ishii et al., 2005; Allmann and Shearer, 2007; Pulido et al.,
2008), it is possible to obtain the image of source rupture very quickly due to lower computational demands.
The method provides a stable estimate of the fault rupture velocity in addition to providing the location of
asperities. The BPM directly back projects the energy observed (envelope of the seismogram) at each near-
source seismic station into an image of the earthquake rupture. The method requires the calculation of
isochrone times for each station and for grid points distributed within the source fault plane. The
computational time can be reduced further by keeping a database of theoretical travel times (which is are
required to calculate isochrones) between the various stations and grid points within the potential faults. Our
main goal is to implement BPM with the Quake-Catcher Network (QCN) to provide near-real-time estimates of
the earthquake source process. The QCN is a seismic network that connects computers with internal or
external Microelectromechanical System (MEMS) sensors using over the Internet. The QCN, with the potential
for thousands to hundreds of thousands of sensors, can provide the ample number of seismograms that are
required for the successful application of the BPM method. We are currently studying the potential
applicability of rapid application of the BPM, by applying the method to the 2004 M 6.0 Parkfield earthquake
that has the required density of near-field data. Once the method has been applied successfully to the
Parkfield earthquake, we will check its applicability with seismograms recorded by QCN networks. Eventually,
we hope to develop a set of automatic algorithms to implement the BPM in real time with the QCN network.
http://qcn.ucr.edu
G43A-0665
The Earthscope USArray Array Network Facility (ANF): Metadata, Network and Data Monitoring, Quality Assurance as We Start to Roll
The Array Network Facility (ANF) for the Earthscope USArray Transportable Array seismic network is responsible for: the delivery of all 400+ Transportable Array stations to the IRIS Data Management Center, collection of regional network stations which contribute data to the Transportable Array; station command and control; verification and distribution of metadata (~560 current and former TA stations as of September 2008); providing interfaces for personnel at the Array Operations Facility (AOF) to access state of health information; and quality control for all data. To meet these goals, we use BRTT's Antelope software package to: facilitate data collection and transfer; generate and merge station metadata; monitor real-time datalogger state-of-health; and review seismic events. Weekly transfers of dataless SEED and Virtual Network Definitions (VNDs) are simplified by the use of ORB transfer technologies at the ANF and receiver end points. Extensions to the Antelope software package have been contributed to help with data center operations. Additional software packages including Dartware's InterMapper network monitoring application and Round Robin Database Tool monitor and report on hardware or communications failures. The on-going quality control process includes: 1) automatic event processing followed by daily analyst review associating arrivals against available regional network bulletins (36000+ events and 1.9 million picks); 2) review of clock quality and error; 3) review of number of mass recenters; 4) review of percent of time any of the three mass positions are out of range; 5) alarms upon datalogger reboots; 6) alarms upon active pumps; and 7) review of calibration signals at each station upon installation and prior to removal. Much of this information is available via interactive online tools at the ANF website (http://anf.ucsd.edu).
G43A-0666
Managing a Torrent of Diverse Low-Latency Data Streams at the IRIS DMC
In 2003 the IRIS Data Management Center (DMC) started a major effort to collect low-latency streaming data. Within a few years this near real-time data collection system, called the Buffer of Uniform Data (BUD), became and remains the largest source of data for the DMC's archive. At the current streaming data collection rate of 25 Gigabytes per day these volumes represent significant data flow by most standards. The BUD system collects data from over 70 suppliers using 4 different protocols. More than 16,000 discrete channels of data are received from over 1,700 stations continuously. The bulk of these data are seismic but also include strong-motion (engineering), pressure, strain, gravimeter and infrasound data. The primary reason for the DMC to collect low-latency data is to streamline the data archiving process. Continuously arriving data can be archived in a reliable, automated way that uses resources efficiently. The task for the data submitter is likewise improved. An obvious secondary reason to collect low-latency data is to provide our users with rapid access to data. Users can access data directly from our BUD data collection buffers and earthquake-related products are bundled and available within hours of each large earthquake. Additionally, all open data streams are exported as low-latency data streams using a single protocol (SeedLink) and uniform format (Mini-SEED). The latency introduced by our collection system for our data export streams is dependent on the sampling rate of the data with high-rate data (e.g. 40+ Hz) having effectively no delay added and increasing delay added for lower rate streams. A significant advantage of user access to the BUD via any of the access mechanisms is the unified data format. Another opportunity advantaged by collecting streaming data is the ability to process data as it arrives. The DMC's Quality Analysis Control Kit (QUACK) runs multiple measurements on all open data streams collected. These measurements include simple calculations such as data completeness and signal RMS to more advanced spectral analyzes to measure background noise levels. The results are available via a flexible web interface and are useful not only for the DMC to monitor data flow but also for station operators interested in quality assurance measurements they may not be doing themselves and to allow researchers to review characteristics of desired data channels. Making these results available for the low latency data allows early identification of data problems. In summary, the DMC's low-latency data collection system facilitates the addition of a large volume of data into the archive and is the foundation for low-latency data access for users as well as data stream monitoring and characterization.
G43A-0667
Collecting and Using Low Latency Data at Berkeley Seismological Laboratory
Northern California and the San Francisco Bay Area are among the US regions that combine high earthquake hazard and high population density. To rapidly and reliably monitor tectonic movement and develop an understanding of fault dynamics, measurements must cover a range of scales in time (0.1 s to years), space (mms to 100s of km) and displacement (microns to 10s of m). With these goals in mind, Berkeley Seismological Laboratory (BSL) continuously collects a wide variety of data at low latencies from seismic through geodetic, strain and electromagnetic instrumentation with sampling rates spanning 0.001 sps to 500 sps. Data from broadband seismometers and accelerometers, generally with latencies of less than 10 s, contribute to real time earthquake monitoring in Northern California including rapid assessments of source (moment tensor and finite fault) and shaking (ShakeMap). The BSL is also currently operating a real time system in test mode, using these data for earthquake early warning (ElarmS). Data from these instruments are also used for research on earthquake sources and scaling, fault-related tremor and studies of local, regional and global velocity structure. Low latency GPS data can complement seismic data, contributing robust real time continuous information especially for large earthquakes, and can potentially contribute to early warning. GPS-derived static deformation gives an independent estimate of fault orientation and dimensions, scalar seismic moment and magnitude. It also can extend the upper limits of a strong motion network to include the displacements of tens of meters expected in large and great earthquakes, and in the near field is less likely to be clipped during large movements. In an active tectonic context such as Northern California, low latency is important for data transmission, but also for reliability. At the BSL we are committed to using telemetry that is as robust as possible and often have more than one telemetry path to ensure data delivery. Here we present latency reports for our datasets, results from our work to incorporate GPS data into real time processing, and place these results in the framework of the active collaboration with our partners.
G43A-0668
The Self-Organising Seismic Early Warning Information Network
The Self-Organizing Seismic Early Warning Information Network (SOSEWIN) represents a new approach for Earthquake Early Warning Systems (EEWS), consisting in taking advantage of novel wireless communications technologies. It also sets out to overcome problems of insufficient node density, which typically affects present existing early warning systems, by having the SOSEWIN seismological sensing units being comprised of low-cost components (generally bought "off-the-shelf"), with each unit initially costing 100's of Euros, in contrast to 1,000's to 10,000's for standard seismological stations. The reduced sensitivity of the new sensing units arising from the use of lower-cost components will be compensated by the network's density, which in the future is expected to number 100's to 1000's over areas served currently by the order of 10's of standard stations. The robustness, independence of infrastructure, spontaneous extensibility and a self-healing/self-organizing character in the event of failing sensors during an earthquake makes SOSEWIN particularly useful for urban areas. Moreover, in the post-event time frame, negligible assumptions or interpolations would be necessary for assessing the strong ground shaking and earthquake intensities. In SOSEWIN, the ground motion is continuously monitored by conventional accelerometers (3-component) and geophones and analyzed using robust signal analysis methods by each sensing node of the network. The incoming signals are pre-processed by bandpass filtering and the detection processing is performed using an automatic STA/LTA trigger algorithm. Signal attributes are iteratively estimated from the P-wave part of the recordings (e.g. PGAP, PGVP, PGDP, Arias Intensity and Cumulative Absolute Velocity) to determine if the earthquake is of sufficient magnitude to be of concern to issue a system alarm. Differently from most existing EEWS where the alarming system relies on estimates provided by only a few seismic stations, the SOSEWIN is specifically designed to take advantage during the 'event detection' and 'appropriate issuing of alarms' stages of a redundancy of available real-time ground motion information, thanks to the dense wireless mesh network. All of these strategies are devoted to minimizing the occurrence of false alarms while maximizing the early warning or lead time. The early warning performance of SOSEWIN in terms of its combination of seismological software, hierarchical alarming protocol and routing protocol are currently being tested by simulations. The first deployment of the SOSEWIN was carried out in June 2008, with a network of 20 stations installed in the Ataköy district of Istanbul, Turkey. We present here a report of the first few months of the associated activities, together with what the field experiences have taught us in terms of wireless communication for early warning purposes.
G43A-0669
Wireless technologies for the monitoring of strategic civil infrastructures: an ambient vibration test of the Faith Bridge, Istanbul, Turkey
The monitoring of strategic civil infrastructures to ensure their structural integrity is a task of major importance, especially in earthquake-prone areas. Classical approaches to such monitoring are based on visual inspections and the use of wired systems. While the former has the drawback that the structure is only superficially examined and discontinuously in time, wired systems are relatively expensive and time consuming to install. Today, however, wireless systems represent an advanced, easily installed and operated tool to be used for monitoring purposes, resulting in a wide and interesting range of possible applications. Within the framework of the earthquake early warning projects SAFER (Seismic eArly warning For EuRope) and EDIM (Earthquake Disaster Information systems for the Marmara Sea region, Turkey), new low-cost wireless sensors with the capability to automatically rearrange their communications scheme are being developed. The reduced sensitivity of these sensors, arising from the use of low-cost components, is compensated by the possibility of deploying high-density self-organizing networks performing real-time data acquisition and analysis. Thanks to the developed system's versatility, it has been possible to perform an experimental ambient vibration test with a network of 24 sensors on the Fatih Sultan Mehmet Bridge, Istanbul (Turkey), a gravity-anchored suspension bridge spanning the Bosphorus Strait with distance between its towers of 1090 m. Preliminary analysis of the data has demonstrated that the main modal properties of the bridge can be retrieved, and may therefore be regularly re-evaluated as part of a long-term monitoring program. Using a multi-hop communications technique, data could be exchanged among groups of sensors over distances of a few hundred meters. Thus, the test showed that, although more work is required to optimize the communication parameters, the performance of the network offers encouragement for us to follow this research direction in developing wireless systems for the monitoring of civil infrastructures.
G43A-0670
Bringing High Rate, Low Latency Data From Unimak Island, Alaska
The Plate Boundary Observatory (PBO), part of the NSF-funded EarthScope project, completed the installation of a fourteen GPS stations, eight tiltmeters, one webcam, and one digital broadband seismometer on Unimak Island, Alaska in August, 2008. PBO collaborated with the USGS, who provided engineering support for this project. Combined with the USGS-operated seismic network, the Unimak Island network is a state of the art scientific network. The primary data communications goal of the project was to design and implement a robust data communications network capable of downloading 15-sec daily GPS files and to test the streaming of 1-Hz GPS data at a select set of GPS stations on Unimak Island. As part of the permitting agreement with the landowner, PBO co-located the GPS stations with existing USGS seismic stations. The high-speed radio link deployed allowed the USGS to test the feasibility of broadband seismometer installations on Unimak Island. This collaboration with the USGS was another successful joint operation between PBO and the USGS. The technical and logistical challenges involved in the project as well as some preliminary results of the data communications system will be presented. These challenges include complicated logistics, bad weather, complex network geometries with multiple radio repeaters, long distance RF transmission over water, hardware bandwidth limitations, power limitations, space limitations, as well as working in bear country on an incredibly remote and active volcano.
G43A-0671
The Python Interface to Antelope and Applications
The Antelope Environmental Monitoring System from Boulder Real-Time Technologies, Inc. (http://www.brtt.com) is widely used for acquiring, processing, distributing, and archiving near-real-time monitoring data, especially in seismological networks. We have contributed a new Python interface to the Antelope toolkit, paralleling other commercial and open-source language interfaces in Matlab, PHP, TCL/Tk, and C. The Python programming language (http://www.python.org) is well suited both to scientific computing applications and to interactive web-based applications. In the latter, Python serves as the programming interface through which to connect to standardized open-source frameworks. Community development of these frameworks has advanced in parallel with cross-browser standardization and increasing broadband data transfer rates, making web-based applications the defacto standard for platform-agnostic access to large, heterogeneous datasets. These web-based solutions are starting to mirror some of the capabilities of standard desktop-based applications. We describe the functionality of the new Python interface to Antelope, applications of the interface to the interactive exploration of time-series data on the web using the Twisted open-source framework, and web-based prototype tools developed for the Earthscope Array Network Facility to provide community access to network monitoring and seismic event datasets.
G43A-0672
SafetyNet(TM) -- NPOESS's Low Data Latency Key
A key feature of the National Polar-orbiting Operational Environmental Satellite System (NPOESS) is the Northrop Grumman Space Technology patent-pending data collection architecture known as 'SafetyNetTM'. The centerpiece of SafetyNetTM is the system of fifteen globally-distributed ground receptors developed by Raytheon Company. These receptors or antennae will collect up to five times as much environmental data approximately four times faster than current polar-orbiting weather satellites. Once collected, these data will be forwarded near-instantaneously to U.S. weather centrals via global fiber optic network for processing and production of data records for use in environmental prediction models. Key system design factors: The NPOESS SafetyNetTM architecture provides: Frequent downlinks and maximizes contact duration (>100% margin) at low cost; Downlink bandwidth margin that allows all Stored Mission Data to be down linked to two separate receptors; and Minimal latency impacts from loss of multiple ground receptors. Other notable characteristics of SafetyNetTM: Simple, receive-only, Ka Band receptor design provides autonomous operations; Fifteen locations in ten countries; full-motion to track polar satellites; Reliable and timely collection, delivery and processing of quality data; 75% of NPOESS data products delivered to the US's weather centrals within 15 minutes; the rest in under 30 minutes. Presentation will show: A graphic of SafetyNetTM within the NPOESS program architecture; A depiction of the NPOESS data download scheme; A map of the worldwide SafetyNetTM receptor locations; A graph of the percent of NPOESS Environmental Data Records (EDRs) versus time from observation to delivery; A example chart of the NPOESS data downlink patterns to SafetyNetTM receptors; Example photos of SafetyNetTM receptor antennae and radomes.