Exploiting the Potential of Space-Based Geodesy to Aid Rapid Assessment and Response to Large Earthquakes and Related Phenomena

Transcription

1 Exploiting the Potential of Space-Based Geodesy to Aid Rapid Assessment and Response to Large Earthquakes and Related Phenomena Mark Simons, Caltech Susan Owen, JPL Paul Rosen, JPL Frank Webb, JPL

2 Exploiting the Potential of Space-Based Geodesy to Aid Rapid Assessment and Response to Large Earthquakes and Related Phenomena What is geodesy? Outline The Advanced Rapid Image & Analysis (ARIA) Center Demonstrate different capabilities of space geodesy (InSAR & GPS) Examples: Earthquakes large and small, tsunamis, volcanoes, Where do we stand today vs. where we could be in 3 to 5 years We emphasize the combination of data with modern geophysical modeling Caveat: Currently, all examples are serendipitous

3 What is Geodesy? Geodesy is the measurement and representation of the Earth, in a three-dimensional time-varying space.

4 1999 Mw 7.1 Hector Mine (CA) Contours of ground movement ~2.5 cm per contour 20 m per pixel

5 A Busy Few Years Magnitude M w 8.0 Pisco, Peru 2001 M w 8.4 Arequipa, Peru 1995 M w 8.1 Antofagasta, Chile 2006 M w 8.3 Kuril Islands 2004 M w 9.3 Aceh, Sumatra 2005 M w 8.7 Nias, Sumatra 2003 M w 8.1 Tokachi-Oki, Japan 2007 M w 8.1 Solomon Islands 2009 M w 8.1 Somoan Islands

6 2009 MW 7.6 PADANG, SUMATRA As reported on 2 days after the event Amelia Merrick, the operations director for World Vision Indonesia, described the situation as quite devastating. Bridges have gone down, phone lines are in total disrepair, she said. It's difficult for us to assess the situation.

7 The Advanced Rapid Imaging & Analysis (ARIA) Center for Natural Hazards InSAR and GPS Integrating: Space Geodesy Seismology Modeling Science & near real time assessment Potential Partners Earthquake and tsunami source models Seismology 2007 Mw 8.0 Pisco, Peru Subsurface fault slip: 7 meters peak value Preparing for the future

8 WHAT IS ARIA? Natural hazard science/mitigation/assessment/response Leverages world-leading expertise at JPL and Caltech in the advancement and exploitation of geodesy, remote sensing, and geoscience Sanctioned to span research, technology, demonstration & operational activity (a big change!) Start with ARIA-EQ Beyond ARIA-EQ ARIA-Magma (Volcanic activity) ARIA-Fire (During/after fires) ARIA-Flow (Debris flows, deep-seated landslides).

9 THE ARIA VISION Science Earthquake physics Crustal rheology Landslide/debris flow mechanics Volcanology Hydrology Geodesy Multi-spectral imaging Seismic engineering Applied Science Damage assessment from decorrelation Volcano hazards monitoring Nationwide velocity map Tsunami early warning Fire perimeter tracking CTBT verification Landslide tracking Aquifer/oilfield/CO 2 monitoring Levee/coastal/river delta monitoring Decision Tools & Support Rapid detection, classification & description of events or transients with confidence levels Rapid damage assessment Tsunami early warning Provision of public archive of various levels of data and necessary tools for interpretation Integrated product (fusion) support (e.g., fires perimeters, burn history, vegetation types) Triggered deployments or tasking of observation systems

10 NOTIONAL ARIA-EQ RESPONSE

11 LOTS OF EXAMPLES o A quick InSAR tutorial (4 quick slides only ) o 1999 Mw 7.1 Hector Mine, CA EQ ( Haiti-like ) Observations of ground deformation Fault models Damage assessment o 2005 Mw 7.8 Pakistan EQ o Small earthquakes o Volcanic processes o 2007 Mw 8.0 Pisco, Peru EQ Subduction zone Model-based tsunami assessment o 2007 Mw 7.8 Tocopilla, Chile EQ High rate GPS Tsunami early warning

12 Interferometric Synthetic Aperture Radar (InSAR) InSAR provides: 100+ km wide images 20 m spatial resolution Sub-cm sensitivity Reliable global access ERS-2 (ESA)

13 PHASE - A MEASURE OF THE RANGE The phase of the radar signal is the number of cycles of oscillation that the wave executes between the radar and the surface and back again Number of cycles (actually millions) The phase measures distance (range) in wave cycles (averaging over all things within a pixel) One pixel in an image with millions of pixels Interferometry exploits the phase information as follows

14 DIFFERENTIAL INTERFEROMETRY (AKA InSAR ) The interferometric phase is the difference between two observations made from the same location in space but at different times and is proportional to any change in the range of a feature on the surface.

15 INTERFEROMETRIC DECORRELATION Significant sub-pixel rearrangement of reflectors causes the measurement to fall apart, a.k.a. interferometric decorrelation. Maps of decorrelation can be used to automatically detect damage.

16 1999, Mw 7.1 Hector Mine, CA

17 USING PAIRS OF RADAR IMAGES (TOWARDS 3D) InSAR pixel tracking

18 1999 M W 7.1 HECTOR MINE EQ MODEL OF SUBSURFACE COSEISMIC FAULT SLIP Mapview of Coseismic Displacements Significant vertical fault slip Fault slip concentrated at shallow depth (7 to 10 km) Most slip in the northern half

19 DAMAGE DECORRELATION 2/16/2010

20 InSAR Decorrelation 2/16/2010 Detects: Surface faulting Zones of intense shaking Damage Future Degree of damage Emergency response

21 1999 MW 7.1 HECTOR MINE EQ Decorrelation InSAR displacement Pixel Offsets Seismicity Seismic waveforms GPS All lead to high quality models of subsurface fault slip, predicted ground motions, and estimates of where damage occurred.

22 The Mw 7.6, Kashmir Earthquake of October 8, 2005 Feature tracking using optical imagery, e.g., ASTER, SPOT,

23 The Mw 7.6, Kashmir Earthquake of October 8, m

24 SMALL EARTHQUAKES TOO! 30 April 1999 Mw 5.3 Zagros, Iran 2/16/ June 1992 Mw 5.4 Little Skull Mountain, NV GCMT/ISC (seismology alone) Horizontal location error (67 km /7 km) Depth error (41 km /31 km) Lohman & Simons, 2005 Lohman et al., 2002

25 VOLCANO ACTIVITY Global monitoring of subsurface magma migration before and during eruptive activity to minimize exposure and aid response

26 VOLCANO ACTIVITY

27 TEMPORAL VARIABILITY OF MAGMA CHAMBER PRESSURIZATION magma chamber volume change vs. time

28 1995 Mw Mw Mw Mw Mw Mw 7.8* 2007 Mw 7.7** 2007 Mw 8.0

29 AUGUST 15, 2007 MW 8.0 PISCO (PERU) Sladen et al., 2010

30 Integrating geodesy and seismology 2/16/ , Mw 8 Pisco (Peru)

31 Earthquake slip models constrained by different data 2/16/2010 All fit seismic data equally well Yet, the spatial distribution of fault slip is different -- different tsunami prediction

32 PREDICTED VERTICAL MOTION OF THE SURFACE

33 Testing tsunami model with deep-ocean pressure (DART) sensors. Only models with geodetic data gets both amplitude and timing right.

34 TSUNAMI: INUNDATION PROFILE

35 A prototype tsunami early warning system CANTA GPS network Collaboration with Chilean and Peruvian scientists Autonomous open field sites 5 Hz GPS receivers Some telemetered -- most manually downloaded

36 Nov. 14, Mw 7.8 Tocopilla, Chile Circles: GPS sites

37 Mw 7.8 Tocopilla, Chile Nov. 14, 2007 Mt Chajnantor 5011 m (E. of Salar de Atacama)

38 Detection Near field record sees evolution of fault slip Detection based on raw and/or filtered data (displacements). Reference to recent location (e.g., an hour ago) receivers earthquake GPS time series

39 Nov 14, 2007 Tocopilla Mw 7.8 Our goal: This (or better) in near real time JRGN (tip of Mejillones Peninsula) Preliminary 1 sec GPS solutions (not filtered) From Susan Owen Up East North

40 2/16/2010

41 GPS-based tsunami early warning Rely on near field signal: - Local GPS stays on scale - Local Automatic response applications - Regional Maximize warning time - Global Useful source estimation in minutes Technical comments: - < 10 sec latency < EQ DT (JPL) - Near real time accuracy < 10 cm (JPL) - Look for consistent signal on nearby sites - GPS network should have 50 to 100 km spacing - System health monitoring: Same data flow used by scientists Telemeter raw data Process centrally 1st Attempt Failed

42 Tsunami Warning Soapbox For early warning at large distances, existing seismology can do a great job takes advantage of a large existing global network Near field warning is the realm of real-time GPS (it stays on scale). Event detection is trivial based on the raw time GPS series (if several neighboring GPS sites move more than ~10 cm towards the ocean, then yell wolf). The question is quantification. How much can we say and how quickly can we say it? In other words, how much can we add relative to what tsunami centers already do?

43 GPS OPEN FIELD TELEMETRY ISSUES Manual downloads No routine health of network information Intermittent and logistical complexity Rely on onboard storage for high rate data Can t use for early warning Local radio/internet Generally intermittent (multiple points of failure) Guaranteed to fail during a big event (it did) Putre, Chile

44 GPS OPEN FIELD TELEMETRY ISSUES Only solution: Robust satellite telemetry

45 WHAT DOES THE FUTURE LOOK LIKE?

46 HAITI M W 7.0 How quickly can we produce this?

47 DATA LATENCY - SATELLITE INSAR Last 15 yrs to next couple yrs - Sporadic image acquisitions ESA: ERS-1, ERS-2 & Envisat (C-band) JAXA: JERS-1 & ALOS (L-band) CSA: Radarsat (C-band) ESA: Sentinel 1a/b (C-band) 2012/2013 launch Staggered 12 days orbits => 6 day pairs Low latency data availability (in the cloud) JAXA: ALOS2 (L-band) 2013 launch 14 day pairs NASA: DESDynI (L-band) - < 2017 launch 8 day pairs Low latency data availability (in the cloud) Combine ascending / descending orbits about 1 day access.

48 EXAMPLE: MEAN ACCESS TIME FOR DESDynI Mean Access Time (Day) Steven Hu, JPL

49 SUMMARY High rate, near real time GPS is here with the potential to impact EQ source modeling and enable tsunami early warning but we need robust telemetry for open field installations (This will not happen if it depends only on scientists) Satellite-based geodetic imagery will soon be available with ~1 day latency we need to invest in the analysis infrastructure to have the right products (i.e., damage maps, surface displacement maps, fault trace maps, subsurface slip models, ground shaking predictions) ready for the appropriate end users The best approach is to merge next generation data with next generation geophysical models

50 Hours since EQ Hour 1 Hours 2-12 Hours Day 2 HAITI TIMELINE ACTUAL AND POTENTIAL CNN Report A magnitude 7.0 earthquake struck southern Haiti, the U.S. Geological Survey reported. A tsunami watch is in effect for Haiti, Cuba, the Bahamas and the Dominican Republic. First estimate of epicenter Eyewitnesses report heavy damage and bodies in the streets of the capital. First reports of damage from eyewitnesses. Reconnaissance flights to go over Haiti soon, says U.S. State Dept spokesman. "Port-au-Prince is flattened... more than 100,000 are dead," Haitian consul general to U.N. First public confirmation that the city is heavily damaged and that key health and safety infrastructure has been lost. A large aftershock The World Bank promised $100 million in emergency funding. How was the need assessed? Red Cross estimates one in three Haitians, about 3 million people, affected. "We have enormous numbers of people trapped under the rubble," United Nations' staff. First estimate of the number of people affected. UN requests help. Extent of damage/need still uncertain. The pier used for delivery of cargo to Port-au-Prince by ship was "completely compromised." Relief supplies prevented from being off loaded. 2/16/2010 Potential Future ARIA Center Response Report of EQ received from NEIC, triggering automated download of seismic data, search for available geodetic data Improved seismic model, combined with GPS (if available) within an hour. Improved estimate of EQ magnitude, location, faulting style supplied to response agencies and tsunami warning centers for better ground shaking/damage assessment. Update GPS displacements and EQ model. Send updated model to response agencies. Generate interferometric displacement and decorrelation images of EQ. Continue to update GPS displacement estimates. Update EQ model combining InSAR, GPS, and seismic data. Continue to transmit updates to response agencies. Send damage detection map to response agencies, providing crucial information on extent of damage over region. For large aftershocks, determine displacements; include aftershocks in updated kinematic and dynamic models. Collect, analyze optical imagery, analyze and provide to response agencies for improved assessment of damage. Generate additional displacement observations as more data become available, incorporate new data into damage detection product, and use displacement maps to improve EQ models. Deliver updated products to response agencies and science community.

51 THANK YOU