Upload
kaylyn-romer
View
216
Download
0
Tags:
Embed Size (px)
Citation preview
On the potential of satellite TIR surveys for a Dynamic Assessment of (short-term) Seismic Risk: some examples from the EU-FP7 PRE-EARTHQUAKES Project
Valerio Tramutoli1,2
1 University of Basilicata, Potenza – Italy ([email protected])
2 Institute of Methodologies of Environmental Analysis, CNR, Tito Scalo – Potenza – Italy
SCEC CSEP Workshop on Testing External Forecasts and PredictionsLos Angeles 7-8 May 2013
TIR SIGNAL
...
)(
0
dhh
hhTBTBR
sath
s
surfaceTS: surface temperature eDl: spectral emissivity
NOAA-16 July, 22, 2001 01GMTAVHRR Channel 4
total atmospheric transmittance and profile which depends on physical (mainly T(h) temperature profile) and chemical (mainly H2O, CO2, NH4) properties of the atmosphere
atmosphere
Data Analysis: the noise ! (natural/observational)
surface
atmospere
Atm
ospheric
transmittance
Surface
temperature
Spec
tral
emiss
ivity
orog
raph
y
Atmospheric
temperature
profile
Atmospheric humidity
profile
Time o
f day
Season
Satellite view angleSatellite
spatial
resolution
Satellite TIR
signalTIR signal is strongly variable depending on the observation time t and place r.
TIR Anomaly Monitoring by RST (Robust Satellite Technique):
the RETIRA (Robust Estimator of TIR Anomalies) index
(Tramutoli et al., RSE, 2005)
t 12
5
17
y x reducing site effects
Signal time-average µΔT(x,y) and standard deviation ΔT(x,y)
are computed at the pixel level in similar observational conditions (same month of the year, same time slot, etc.)
reducing year-to-year variability and seasonal-drift effects
The local signal excess T(x,y,τ) = T(x,y,τ) - <T(τ)> (compared with the spatial average on the scene) is the considered signal instead of its absolute local value T(x,y,τ)
• Validation/confutation approach always applied
∆T(x,y,τ) ),(
),(),,(
yx
yxyxT
T
T
V(x,y,τ) = T(x,y,τ); T(x,y,τ) = T(x,y,τ) - < T(τ)>
space-time persistence requiredknown spurious effects discarded ( Filizzola et al., 2004, Aliano et al., 2008, Genzano et al., 2009)
RST derives from RAT (Robust AVHRR Approach)
(Tramutoli , 1998)
A posteriori Validation/Confutation Analyses
(5.7 – 9.0) MAGNITUDE EQs EVENT TECHNIQUE
23 November 1980, Irpinia-Basilicata-Italy, Ms=6.9
AVHRR -DTIR (Tramutoli et al., Annals of Geophysics, 2001)
23 November 1980, Irpinia-Basilicata-Italy, Ms=6.9
AVHRR - DLST (Di Bello et al., Annals of Geophysics, 2004)
26 September 1997, Umbria, Italy Ms=5.9 to 6.4
METEOSAT - DTIR(Aliano et al., Annals of Geophysics, 2008)
3-7-12-14 October 1997, Umbria, Italy Ms=5.7 max
METEOSAT - DTIR(Aliano et al., Annals of Geophysics, 2008)
17 August 1999, Kocaeli-Izmit, Turkey, Ms=7,4 METEOSAT - TIR(Aliano et al., Annals of Geophysics, 2008)
17 August 1999, Kocaeli-Izmit, Turkey, Ms=7,4 METEOSAT - DTIR (Tramutoli et al., Rem. Sens. Env. 2005)
7 September 1999 Athens Ms=5.9 AVHRR - DLST (Filizzola et al., Phys. Chem. Earth, 2004)
7 September 1999 Athens Ms=5.9 METEOSAT - DTIR (Filizzola et al., Phys. Chem. Earth, 2004)
16 October 1999, Hector Mine, CA, Ms=7,4 GOES - DTIR (Aliano et al., Annals of Geophysics, 2008)
26 January 2001 Gujarat, India Ms=7.7 METEOSAT - DTIR (Genzano et al., Tectonophysics, 2006)
21 May 2003 Zemmouri, Algery Ms=6.9 METEOSAT - DTIR (Aliano et al., IEEE, Multi-Temp, 2007)
6 April 2009, Abruzzo, Italy M=6.3 AVHRR - DTIR, MODIS - DTIR, METEOSAT - DTIR (NHESS,Genzano et al., 2009, Pergola et al., 2010, Lisi et
al., 2010)
11 March, 2011 Tohoku, Japan M=9 MTG - DTIR (Genzano et al., AGU, 2011)
The method has been independently evaluated by two projects funded by two National Space Agencies (NASA and DLR)
Greece-Turkey 1998-2004
AVHRR 6 YEARS California 2000-2006MODIS 7 YEARS
7
V.Tramutoli__ EMSEV 2010 ___Chapman University, Orange, CA, USA ___([email protected])
Moving to multiparametric observations (Abruzzo-L’Aquila April 6th 2009 EQ)
TIR anomaliesMODIS-AVHRR-SEVIRI 30/3/09Number of EQ
January 1 up to April 6, 2009
0 10 20 30 40 50 60 70
-5
0
5
4
2
resi
dual
days since march 31, 2009
Mag
nitu
de
CO2 fluxesMartinelli, 2009
30/3/09
Uranium groundwaterPlastino et al 2010
Genzano et al; NHESS; 2009Lisi et al; NHESS; 2010
Pergola et al; NHESS; 2010Vp/Vs anomalies
Lucente et al, (Geology, 2010)
G. Papadopoulos et al., 2010
VLF radio anomalies
VLF radio anomaliesRozhnoi et al 2009
De Santis et al., 2010
30/3/09
TEC anomalies Akhoondzadehet al., 2010
8
Laboratoire the Physique et de Chimie de l’Environment and de l’Espace - CNRS
Partners
WD IZMIRAN FIAG
UNIBAS
DLR
TUBITAK MAM
RSS
GSI
LPC2/CNRS
CHAPNOA
(increasing through Networking Membership)
National Observatory of Athens
ChapmanUniversity
Strategy
inde
pend
ent
o
bser
vatio
ns
Integration tool:Pre-Earthquakes Geoportal
(PEG)
1. Learning
Off-line integration on
past events over 3 main
testing areas/events
2. Apply in Real Time: PRIME (Pre-earthquakes Real-time Integration and Monitoring Exercise)
Real time integration
over 2 selected wide areas
Jul– Nov 2012
and tools
From “a posteriori” to real time validation/confutation
PRIME The Pre-earthquakes Real-time Integration
and Monitoring Experiment(July-November 2012)
18-JUL-2012
DECISION: To locally look to the data, Blue (Only Local) Alert Level
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
19-JUL-2012
DECISION:, REQUEST OF ATTENTION BY
PARTNERS !Yellow Alert Level
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
20-JUL-2012
DECISION:, MOVE TO RED ALERT
20-JUL-2012
22 July 2012 09:26:02 UTCM=5
ANDIRIN – KHARANMANMARA event occurred(4 days after the first anomaly observed on July 18)
PRIME results
Example 2K.Maras-Pazarcik event
(16/10/2012 M 4.6)
The game of responsibility a real time experimentPresented at the EMSEV 2012 Conference
Gotemba Japan 3 October 10:45 LT
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
28-SEPTEMBER-2012non persistent anomalies
DECISION: To Locally look to the data, Blue (Only Local) Alert Level
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
29-SEPTEMBER-2012 2 persistences close to faults
DECISION:, REQUEST OF ATTENTION BY
PARTNERS !Yellow Alert Level
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
DECISION:, MOVE TO RED ALERT
30-SEPTEMBER-20121 persistent high intensity close to the fault
01-OCTOBER-2012no anomalies (but the alert status should be
mantained for the next 6-7 days)
OCTOBER 1, 2012 (18:17 Rome time)
Why ?Color code
Total score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
DECISION:, MOVE TO RED ALERT
WHY ?
20 July 2012 (2 days before the M5 EQ)
Same shape, same place than in the previously predicted event (22 July 2012, M=5)
22 July 2012 09:26:02 UTCM=5
29 September 2012 30 September 2012
WHY ?
29 September 2012 30 September 2012Anomalies (again) following the main fault systems
OCTOBER 1, 2012 (19:43 Moscow time)PRE-EARTHQUAKES collaboration
(FIAG, Sergey Pulinets)
PRE-EARTHQUAKES final review meeting - Brussels - 15 March 2013
OTHER AVAILABLE INFOMedium term forecast (e.g. M8+, Kossobokov)
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
Color codeTotal score
Actions
Look to the data (local)
1
Look to your data (all network)
2Triggering the attention of
all partners to carefully control their data
Alert: EQ Possible 3
All partners are requested to provide their data until
green light is given.
End-Users are informed by the Local Reference
Partner
Alert: EQ Very Likely
≥4
All partners are requested to continue to provide their
data until green light is given. End-users are solicited by the Local Reference Partner to carefully consider the
general situation
(asking for other independent information)
End of Alert ≤1Come back to the original
activity status
OCTOBER 2, 2012 (13:29 Berlin time)
PRE-EARTHQUAKES collaboration (DLR, Norbert Jakowski)
28 09 2012 14:25:00 UT
28 09 2012 16:25:00 UT 28 09 2012 17:05:00 UT
29 09 2012 14:05:00 UT 29 09 2012 16:45:00 UT
More than 10 hours/day of anomaly on Sept 28-29
(see Tiger Lee speach)continuing 30 sept and Oct 1 but with
geomagnetic storms
VIOLET ALERT
Participants to the EMSEV 2012 Conference Gotemba (Japan) 3 October 2012
2 weeks later EMSEV presentation 16 October 2012
2012-10-16 10:25:05 UTC
ML 4,5Depth 39.4 Km
2012-10-16 01:16:02 UTC
ML 4,6Depth 9.2 Km
Conclusions (? only 4 months of test )
During PRIME experiment significantly persistent TIR anomalies were observed only in few cases (very often in apparent relation with earthquakes occurred within 2 weeks later).
Integration with ancillary information and/or observations allowed to increase the number of pre-alerted events
Very few earthquakes with M>4 (not necessarily the most important) have been pre-alerted compared with their total number (>60, clouds coverage playing the major preventing role).
The repetition of pre-seismic TIR anomalies with similar characteristic in the same place strongly increases reliability of the alerts (shared with the EMSEV community 2 weeks before)
The italian (DPC-INGV-S3) project on short-term earthquakes prediction
(Coord. Prof. Dario Albarello)
32
1. Describe a typical forecast: Space-time persistent TIR anomalies (not associated to known and verifiable spurious effects) can contribute to increase the alert level in the framework of a DASR System. In association with other information quality of the forecast strongly increase (no apparent relations with magnitude).
2. What area do they cover? variable from 106 up to 106 km2
a. What is the magnitude range? >4 (possible relationships among affected area and M still under study)b. How long is the time period? from 30 to few days before the EQ, co-seismic and after-seismic (until few
days after the EQ) observed (and expected)c. Do you include a probability of an event during the forecast? only Low, High, Very High alert levels are
given always in conjunction with other independent observations. Relative intensity of RETIRA index can offer (if time series are sufficiently long to justify a Gaussian-like distribution) an indication on the rarity of the anomalies. Space-time persistence, shapes, spatial relations with other static (e.g. fault system) or dynamic (e.g. seismicity) factors are also considered.
d. Do you include a confidence level in the forecast? see before
3. Describe the process for making a forecast:a. Is it automatic or manual? Automatic for TIR anomalies generation. Partly automatic for the analysis of
spurious effect and space-time persistence.b. What data are used? Thermal Infrared Radiances (10-12 micron) from different satellite sensors onboard
geostationary platforms (like MSG, GOES, MTSAT, etc.) is today preferred. Passive MWs sensors (not yet available) on geostationary satellite should significantly improve negative effects due to cloud presence and distribution (spurious effects)
4. Do you have a preference for which earthquake data should be used to test your forecasts?No
5. Do you have a preferred testing method? Real-time release of (different level of) alerts in the framework of a PRE-EARTHQUAKES like collaboration.
6. What physical hypotheses about earthquake predictability have motivated your research? Like all physical processes earthquakes (and their preparation phases) can be studied, modeled and then predicted (as it is already done at least on the long terms). No scientific proofs exists demonstrating an intrinsic EQ unpredictability. Main problem is, presently the accuracy of predictions and the poorness of the observation systems. .
Forecast Summary QuestionsTIR anomalies by RST approach and RETIRA index (Tramutoli et al., 2005)
33
RAT/RST people
Thanks