What Happens.... - PowerPoint PPT Presentation

1 / 19
About This Presentation
Title:

What Happens....

Description:

Nineteen significant earthquakes (blue circles) have occurred in Central or Southern California. ... Major earthquakes m 6 tend to occur when Intensity ... – PowerPoint PPT presentation

Number of Views:147
Avg rating:3.0/5.0
Slides: 20
Provided by: cse70
Category:

less

Transcript and Presenter's Notes

Title: What Happens....


1
What Happens.... When You Include the Test in The
Forecast?? Subtitles The Importance of a Good
Test ...and... The Need for Ensembles of Forecasts
John Rundle, James Holliday, Kristy Tiampo,
Jordan Van Aalsburg, Don Turcotte, Kazu Nanjo, CC
Chen, W Klein With thanks for many educational
conversations with Bernard Minster, Tom Jordan,
Dave Bowman, Charlie Sammis, Ned Field, V
Keilis-Borok, V Kossobokov, P. Shebelin, D.
Sornette, A. Helmstetter, Mike Blanpied, Dave
Jackson, Bill Ellsworth, Jim Dieterich, Jeremy
Zechar...and many others
2
Status of the Real Time Earthquake Forecast
Experiment
Nineteen significant earthquakes (blue circles)
have occurred in Central or Southern California.
Margin of error of the anomalies is /- 11 km
Data from S. CA. and N. CA catalogs After the
work was completed 1. Big Bear I, M 5.1, Feb
10, 2001 2. Coso, M 5.1, July 17, 2001 After
the paper was in press ( September 1, 2001 )
3. Anza I, M 5.1, Oct 31, 2001 After the paper
was published ( February 19, 2002 ) 4. Baja I,
M 5.7, Feb 22, 2002 5. Gilroy, M4.9 - 5.1,
May 13, 2002 6. Big Bear II, M5.4, Feb 22,
2003 7. San Simeon, M 6.5, Dec 22, 2003
8. San Clemente Island, M 5.2, June 15, 2004
9. Bodie I, M5.5, Sept. 18, 2004 10. Bodie II,
M5.4, Sept. 18, 2004 11. Parkfield I, M 6.0,
Sept. 28, 2004 12. Parkfield II, M 5.2,
Sept. 29, 2004 13. Arvin, M 5.0, Sept. 29,
2004 14. Parkfield III, M 5.0, Sept. 30,
2004 15. Wheeler Ridge, M 5.2, April 16,
2005 16. Anza II, M 5.2, June 12, 2005 17.
Yucaipa, M 4.9 - 5.2, June 16, 2005 18.
Obsidian Butte, M 5.1, Sept. 2, 2005 19.
Baja II, M 5.4, May 23, 2006 Note This
original forecast was made using both the full
Southern California catalog plus the full
Northern California catalog. The S. Calif catalog
was used south of lattitude 36o, and the N.
Calif. catalog was used north of 36o . No
corrections were applied for the different event
statistics in the two catalogs. Green triangles
mark locations of large earthquakes (M ? 5.0)
between Jan 1, 1990 Dec 31, 1999.
CL03-2015
Plot of Log10 (Seismic Potential) Increase in
Potential for significant earthquakes, 2000 to
2010
3
Summary
4
Generalization Time Dependent Forecast
Evaluation Using Intensity and Mean Square
Intensity Change Maps for earthquakes m ? mT,
where mT is a threshold magnitude 5 ? mT ? 3
We divide past time into intervals A data
Training Interval, a Forecast Period, and a
Test Interval or Evaluation Interval. Two
magnitude values are important The magnitude of
the forecasted events (here taken to be mF
6.0), and the threshold magnitude mT of the test
events. Length of the Snapshot Interval t
is the current time t2 is determined so
that t t2 is a Gutenberg-Richter Interval
(GRI) corresponding to the forecast magnitude mF
6.0 A GRI contains
events of magnitude m mT
5
Hotspot Maps Based on Intensity and Mean Square
Intensity Change Maps computed with data
beginning in 1932. Intensity change over 13-year
interval prior to date shown. Magnitude
threshold used for evaluation during snapshot
window is mT ? 4.0. ROC Curves are computed for
100 mT ? 4.0 events (GRI for mF 6.0 event)
6
Time-Dependent Forecast Evaluation Oriented area
between the ROC curves is a measure of relative
forecast skill
We break the past data up into a training set,
composed of the intermediate past, and a
forecast set, composed of the recent past. We
forecast the recent past from the intermediate
past using Intensity and Mean Square Change
methods, and compute ROC curves for each. We then
compute the area between the ROC curves.
7
Scale Invariant Gutenberg-Richter Windows Nature
apparently prefers to measure time in
scale-dependent stress release units, i.e,
number of earthquakes having m ? mT
We superpose the curves for ?A(t) over a scale
invariant set of time windows t - t2(mT), where t
is current time. Value of t2(mT) is determined
so that the correct scale-invariant number of
events N(mT) of magnitude m ? mT is included in
each window. Windows are averaged with equal
weight. Averaging over mT? 3.0,5.0 tends to
reduce the noise and random fluctuations, and to
better localize the signal in time, leading to
signal spikes.
8
Results California Between Latitude 32o and
40o Using the Binomial distribution, we compute a
14 chance that the agreement (12 out of 16 sets
of m ? 6 earthquakes in black intervals) is due
to random chance
?A(t)
Whittier Narrows Superstition Hills
Joshua Tree - Cape Mendocino
San Fernando
Imperial Valley
Anza Borrego
Northridge
Hector Mine
San Simeon
Loma Prieta
Mammoth Lakes
Chalfant Valley
Landers-Big Bear
Parkfield
Coalinga
Morgan Hill
Eureka Valley
Time, t
9
Further Dividing California Into North and
South Improves Localization in Time
10
Does the Method Work Elsewhere? Very Early
Results - Central Japan K. Nanjo et al., 2006
11
Summary Integrating a running test into a
forecast can improve performance and can lead to
new avenues of research Ensemble earthquake
forecasts make use of several models to optimize
various aspects of the forecast Patterns and/or
correlations and/or regularities are the means by
which we forecast...so we need to focus on
pattern analysis and recognition, and how to test
these
12
Prediction Hypotheses and Testing Methods
II Motivation
13
Motivation
14
Motivation
15
Tom Jordan on Earthquake Prediction
"I want to be famous. If I can learn how to
predict earthquakes I'll get the Nobel Prize or
something, right?
"The problem is, everybody is looking for a
silver bullet," Jordan says.
From Predicting Big One Eludes Experts By John
Ritter, USA TODAY , 1-24-2006 http//www.usatoday.
com/tech/science/2006-01-24-quake-prediction_x.htm
16
Collaboratory for the Study of Earthquake
Predictability
Which One is Best ??
17
Binary Forecasts Testing and Verification Receiv
er (Relative) Operating Characteristic (ROC)
Diagrams An Application of Signal Detection Theory
Signal Dectection Theory Decision Threshold
Defines Fraction of Probability Map Appearing as
Hotspots
Probability
Decision Threshold
EQ Likely
EQ Unlikely
x (position)
Success
Success is defined if epicenter of large event
falls within the area enclosed by dashed line
18
Enhanced PI Method Applied to California
Earthquakes JR Holliday et al., Nonlin. Proc.
Geophys., (2005) CC Chen et al., Geophys. Res.
Lett., (2005)
We have developed a new enhancement of the
original PI method whose starting point is a
forecast based on the RI map, and then improves
upon it (CC Chen et al., 2005). At right are
maps based on this enhancement corresponding to
the forecast for 2000 - 2010. Details We use
only the top 10 most active sites, and
normalize all time series in the remaining boxes
to have the same statistics. The new algorithm
weights the change maps made using longer time
series more heavily than change maps made using
shorter time series. Here t0 1950, t1
1985, t2 1999. Here we use M 2.8 events to
forecast M 4.8 earthquakes.
ANSS Catalog
19
Comparing the PI Hotspot Map with the USGS
National Seismic Hazard Map Information Content
is Different
USGS National Hazard Map http//earthquake.usgs.go
v/hazmaps/products_data/2002/2002April03/WUS/WUSpg
a500v4.pdf PI Hotspot Map http//hirsute.cse.ucd
avis.edu/rundle/EQ_FORECASTS/CURRENT_SCORECARDS/S
coreCard_Original_Sept_2_2005.pdf
Write a Comment
User Comments (0)
About PowerShow.com