Detection of the 2010 Chile earthquake tsunami from … International Workshop on Remote Sensing for Disaster Management Tokyo, 30 September, 2010 Session: 2010 Chile earthquake Detection

8th International Workshop on Remote Sensing for Disaster ManagementTokyo, 30 September, 2010

Session: 2010 Chile earthquake

Detection of the 2010 Chile earthquake tsunami Detection of the 2010 Chile earthquake tsunami from satellite altimetry

Yutaka HAYASHI1 Norihisa USUI2 Masafumi KAMACHI2 Yutaka HAYASHI1, Norihisa USUI2, Masafumi KAMACHI2, 1 Seismology and Volcanology Research Dep., Meteorological Research Institute,

Japan Meteorological Agency (MRI, JMA)2 Oceanographic Research Dep., MRI, JMA

andShunichi KOSHIMURA3Shunichi KOSHIMURA

3 Graduate School of Engineering, Tohoku Univ.

1 /20

Motivation: Is it possible to monitor/findMotivation: Is it possible to monitor/find tsunami from the space?

Before EarthquakeBefore Earthquake

Earthquake, Propagating TsunamiEarthquake, Propagating Tsunami

2 /20

Contents

1. Introduction, BackgroundsA) Satellite altimetryB) 2004 Indian Ocean tsunami detected by altimetry

2. 2010 Chile earthquake tsunami2. 2010 Chile earthquake tsunami

3 Advanced method to extract tsunami signals from3. Advanced method to extract tsunami signals from satellite altimetry dataA) Methods: Multi-satellite time-spatial interpolationB) Results: 2004 Indian Ocean tsunami profiles along 5 satellite tracks

3 /20

Introduction, Backgrounds

1. Introduction, BackgroundsA) Satellite altimetryB) 2004 Indian Ocean tsunami detected by altimetry

2. 2010 Chile earthquake tsunami2. 2010 Chile earthquake tsunami

3 Advanced method to extract tsunami signals from3. Advanced method to extract tsunami signals from satellite altimetry dataA) Methods: Multi-satellite time-spatial interpolationB) Results: 2004 Indian Ocean tsunami profiles along 5 satellite tracks

4 /20

An altimeter measures sea surface height along satellite tracks

AltimeterMicrowave sensor (C- and K-band)Microwave sensor (C and K band)

Accuracy of observationJason-1, 2 <4.2cm

TOPEX/Poseidon 4.2cm (RMS), ...

Sampling approx every 5-10kmSampling approx. every 5-10km

MissionsOcean monitoring, etc.g,

cycle (day)operation data processingsatellite cycle (day)operation data processingsatellite

List of altimeter-equipped satellites

TOPEX/Poseidon9.9156CNES, NASA

Jason-1

cycle (day)operation, data processingsatellite

TOPEX/Poseidon9.9156CNES, NASA

Jason-1

cycle (day)operation, data processingsatellite

～2006

35.0000ESA, CNESENVISAT17.0506NOAAGFO35.0000ESA, CNESENVISAT17.0506NOAAGFOJason-2 2008～

5 /20

Operating agencies routinely process data to reduce noisesreduce noises

Raw data involve noises caused byyOcean tidesAir pressureOffsets of each satellite reduced by routine processingOffsets of each satelliteGeoid locality, etc.

by operating agencies

(freely distributed products)

SLA ( l l l )

(freely distributed products)Real-time (Jason-2)* 3～6 hours

Near Real-time 1～2 daysSLA (sea level anomaly)

Nonseismic effectsSea currentsTemperature oceanographers

Delayed time 3～6 months

TemperatureWinds … etc.

g pinterest

* Limited to real-time routine use6 /20

Right use of altimetry data- Example: Monitoring El Nino -

Map of monthly mean sea level anomalyMap of monthly mean sea level anomaly

cm

(From website of Japan Meteorological Agency)

cm

7 /20

Profiles of Indian Ocean Tsunami (2004) 2 hours after the main shock were measured by satellites.

SLA diff (SLA) (SLA)

Rough Estimation of tsunami height

SLA difference = (SLA)O.T.+2h. - (SLA)O.T.-10days(SLA : Sea Level Anomaly)

O.T. + 114-123min.

O.T. + 121-130min.

Hirata et al.(2006)Lack of data8 /20

Observation data from satellite altimetry were used for analysis of tsunami mechanism

Important information. Why?Ri h i h i f i h hRich in the information near the trenchLess affected by nonlinear effects

Application (example)Slip distributions by linear inversion of tsunamiSlip distributions by linear inversion of tsunami waveforms

Tsunami source generated slowly (<1km/sec) ? (by Hirata et. al, 2006)

Hirata et. al (2006, EPS)9 /20

2010 Chile earthquake tsunami

1. Introduction, BackgroundsA) Satellite altimetryB) 2004 Indian Ocean tsunami detected by altimetry

2. 2010 Chile earthquake tsunami2. 2010 Chile earthquake tsunami

3 Advanced method to extract tsunami signals from3. Advanced method to extract tsunami signals from satellite altimetry dataA) Methods: Multi-satellite time-spatial interpolationB) Results: 2004 Indian Ocean tsunami profiles along 5 satellite tracks

10 /20

Travel time of the 2010 Chile earthquake tsunami, and altimeter-equipped satellite tracks

11 /20

Sea level anomaly difference calculated by using near real-time data (IGDR) of Jason-2

Near tsunami front (S5-N5) ～0 1m along satellite track (by model of Tohoku Univ )0.1m along satellite track (by model of Tohoku Univ.)Tsunami + shift of oceanographical circulation with long wavelength ?

Snapshot of tsunami propagationSnapshot of tsunami propagation modeled by Tohoku Univ.

12 /20

Advanced method to extract tsunami signals from satellite altimetry data

1. Introduction, BackgroundsA) Satellite altimetryB) 2004 Indian Ocean tsunami detected by altimetry

2. 2010 Chile earthquake tsunami2. 2010 Chile earthquake tsunami

3 Advanced method to extract tsunami signals from3. Advanced method to extract tsunami signals from satellite altimetry data (Hayashi, 2008, JGR) A) Methods: Multi-satellite time-spatial interpolationB) Results: 2004 Indian Ocean tsunami profiles along 5 satellite tracks

13 /20

Can nonseismic effects be reduced? Can the tsunami effect be extracted from SLA change?

Raw data involve noises caused byOcean tidesAir pressure reduced by routine processing

by AVISOOffsets of each satelliteGeoid locality, etc.

by AVISO

(products)SLA

Nonseismic effectsSea currents

(products)

oceanographers Sea currentsTemperatureWinds … etc.

Tsunamiseparate

Seismologists i

g pinterest

Tsunami

Coseismic geoid changeinterest

vanishingly small(by Hayashi et al., 2007, EPS)

An improved tsunami profile has the potential to contribute towards an analysis of huge tsunami

14 /20

To estimate "SLA under the assumption that no tsunami occurred" by multi-satellite time-spatial interpolation

DataAltimetry data from four satellitesSampled data near from reference pointspData not affected by tsunami

are usedR=45×3km- 10×3day < T< +10×3dayare used - 10×3day < T< +10×3day

Typical scale of meso-l dd i hi iscale eddy in this region45km, 10days (Stammer,1997)

All tracks from four altimeter equipped satellites15 /20

Scale parameters R=45km,T=10days are used

Reference heights along ENVISAT track 352Time-spatial interpolation

R=45km, T=5,10,20,30days

T=10days R=15 45 90 180km

Difference of two cycles

T=10days, R=15,45,90,180km

Relation between scale parameters and RMSEsHayashi (2008, JGR)

Typical scale of mesoscale eddy in this region

45km, 10days (Stammer,1997)

Results (products of tsunami profiles after processing) - Example: Jason-1 track 129p g) p 1 129

cycle#109Time-spatial interpolationSimpler wave form

Smaller heightsPeak 0.7 m -> 0.6 m

cycle#109SSHref

interpolation

Peak 0.7 m 0.6 mDefined at more sampling points SSHc#109 – SSHref

cycle#108

O.T. + 114-123min.

Difference of two cycles

cycle#1092004 Indian Ocean Tsunami

SSH – SSH O T + 121 130minSSHc#109 – SSHc#108

Hayashi (2008, JGR)

O.T. + 121-130min.

16 /20

Jason-1, TOPEX/Poseidon, ENVISAT, and GFO successfully detected the Indian Ocean tsunamiy

2004 Indian Ocean Tsunami

121-130min

from mainshockfrom mainshock

201-190min

441-431min 524-544mindetected wave frontHayashi (2008, JGR)17 /20

Comparison of two methods2004 Indian Ocean Tsunami

Difference of two cycles

Multi-satellite time-spatial interpolation

Tracks include tsunami signals 2 (TOPEX/Poseidon 5 (T/P J1 ENVISATTracks include tsunami signals 2 (TOPEX/Poseidon, Jason-1)

5 (T/P, J1, ENVISAT, GFO×2)

Number of valid sampling J-1 356/

412points T/P 88 285Accuracy of data more than 7-10 cm

(RMSE)4-5cm

Necessary period to makereference sea level

2～3 days 5～10 month

Time-spatial interpolation method :Advantage e g ±30days

Time-spatial interpolationg

• accurate• potential of small tsunami detection

Disadvantage

futurepast

1 cycle

e.g. ±30days

DifferenceDifference

• too late to useObservation date used to define reference height, and date of tsunami

1 cycle

18 /20

Summary

The multi-satellite time-spatial interpolation method effectively reducedThe multi satellite time spatial interpolation method effectively reduced the oceanographical backgrounds in the SLA data of 2004 Indian Ocean tsunami.

2010 Chile earthquake tsunami is a bit small to be discussed by “rapid and less-accurate method” (difference of two cycles).By our current best method “slow and accurate” ( lti t llit ti ti lBy our current best method “slow and accurate” (multi-satellite time-spatial

interpolation), it takes many months to extract signals from altimetry data.

More advanced “rapid and accurate” method is necessaryfor the purpose of monitoring tsunamifor the purpose of monitoring tsunami.

19 /20

Thanks for your Attention

20 /20