S32B-01 INVITED
Progress of the New Zealand Earthquake Forecast Testing Center
Steady progress has been made in the last eighteen months in the development of an earthquake forecast testing center for New Zealand. The driving goals of the center have been to: 1) encourage and facilitate the development of new and existing time-dependent forecast models for New Zealand; and 2) examine the models with a series of tests that are appropriate for both the design of the forecast models and also the New Zealand natural lab. The center has begun to explore extensive retrospective testing but ultimately aims to conduct on- going and multi-year testing of models and comparison between models. In order to maintain transparent and widely accepted testing methodologies, we have adopted the testing routines of the Collaboratory for the Study of Earthquake Predictibility (CSEP). Early efforts in the testing center have included improvements to the CSEP testing routines through added computational efficiency. These alternative tests provide equivalent results but reduce the heavy overhead required when running continual testing of multiple models. As with all work done within the center, these additional routines will be available to CSEP and the larger community. Other efforts have included the development of a testing rule-set appropriate to New Zealand and retrospective testing aimed at developing an understanding of characteristics of models implemented within the center.
S32B-02
Earthquake Production by Subduction Zones is Not Linear in Relative Plate Velocity
The ratio of \{long-term-average seismic moment production per unit length of plate boundary\} to \{relative plate velocity\} is determined by the "coupled thickness" of seismogenic lithosphere, and also by elastic moduli and geometric factors that are fairly well known. It is generally assumed that coupled thickness is constant within a given class of plate boundary, such as Bird's [2003, G3]: CCB Continental Convergent Boundary, CRB Continental Rift Boundary, CTF Continental Transform Fault, OCB Oceanic Convergent Boundary, OSR Oceanic Spreading Ridge, OTF Oceanic Transform Fault, or SUB Subduction zone. However, Bird et al. [2002, Geodyn. Ser.] and Bird & Kagan [2004, BSSA] found two exceptions: OSR and OTF both have greater coupled thickness at low relative plate velocities. We test for variation of coupled thickness with relative plate velocity in each of the 7 classes of plate boundary. We use shallow (<70 km) earthquakes from the Harvard CMT catalog, 1982.01.01-2007.03.31, above magnitude MW threshold of 5.51 or 5.66. In order to reduce the influence of aftershock swarms, we estimate the probability of independence of each earthquake according to the likelihood stochastic declustering method of Kagan & Jackson [1991; GJI] and use this as a weight. We use the algorithm of Bird & Kagan [2004, BSSA] to assign 95% of shallow earthquakes to plate boundary steps and plate boundary classes, rejecting all earthquakes that fall into one of the 13 orogens of Bird [2003, G3]. We order the plate-boundary steps outside orogens in each class by relative plate velocity according to the PB2002 model of Bird [2003]. Then, we plot cumulative earthquake count as a function of cumulative model tectonic moment (assuming constant coupled thickness and other parameters within each plate boundary class). The null hypothesis is a linear relation; we use 2 measures (Kolmogorov-Smirnov, and Cramer-von Mises) to quantify departures from this line. We use 10,000 simulations of each class with random Poissonian seismicity in each plate boundary step (with expectations based on the tectonic model) to assess the significance of the measures obtained. Subduction zones have velocity-dependent coupled thickness: P < 0.001 for the null hypothesis. Subduction zones with relative plate velocity <67 mm/a (which would comprise 35% of the model tectonic moment rate, in the null hypothesis) actually produce only 20% of the global subduction zone earthquakes (outside orogens), and thus have a coupled thickness about half that of faster subduction zones (if corner magnitude and spectral slope are constant). This result contradicts the uniform coupling of subduction zones inferred by Kreemer et al. [2002, Geodyn. Ser.]; the difference may be due to their exclusion of several slow subduction zones including Aegean, Cascadia, New Zealand, Caribbean, and South Shetland. Continental CCBs show a similarly strong relation (P < 0.001), with an increase in coupled thickness when velocity exceeds 25 mm/a. OSRs show coupled thickness declining with velocity, as in previous studies. OTFs and OCBs give complex results with significant variations (P < 0.01; P < 0.05) that are not easy to interpret. For CRBs and CTFs we do not reject the null hypothesis of constant coupled thickness. http://peterbird.name
S32B-03 INVITED
Deep Tremor as a Focus of Earthquake Forecasting Research
The discovery of deep tremor has opened a new window into fault-zone processes and offers new possibilities for earthquake forecasting research. Locales where deep tremor has been extensively documented to date include the Nankai Trough, Cascadia, and Central California, which are all areas with considerable seismic potential. Deep tremor is observed to accompany slow slip, such that the combined phenomena have come to be known as episodic tremor and slip (ETS). The emerging view is that deep tremor is the seismic signature of slow slip that occurs episodically on the down-dip extension of seismogenic faults. The amount of slip in these slow slip episodes is approximately equal to the amount of slip that would have been accumulated during the time since the last ETS episode. This suggests that the tremor source may outline the down-dip edge of the locked, seismogenic zone, with the down-dip transition zone locked between ETS episodes, and relaxed during them. Thus, on faults on which tremor occurs, it should be possible to map precisely the down-dip extent of the locked zone, which is often a major source of uncertainty in assessing earthquake potential. Tremor may also prove useful for time-dependent earthquake forecasting. Due to it's strategic location, and episodic nature, deep episodic transients will accelerate stress accumulation on adjacent, shallower, locked portions of faults, which should in turn increase the likelihood of a large earthquake. Thus, tremor is a potentially important forecasting tool, though it might lead to only modest probability gains. With frequent tremor episodes, and infrequent large earthquakes, it will be difficult to test this hypothesis. A more fruitful approach may be to monitor the variation of micro-earthquake activity with respect to deep tremor, and perhaps to examine historical seismic data retrospectively for seismic tremor near the time of large earthquakes. The potential promise of this newly discovered phenomenon argues for enhanced monitoring in areas where tremor is known to occur, and in places where it is likely to occur, based on our understanding of the factors that control tremor occurrence.
S32B-04
Earthquake forecasting by the time series analyses of earthquake source parameters
We collect earthquakes (EQfs) from a focus catalog of Japan Metrological Agency (JMA) for a small mesh region of about 5 degrees and a large region of about 30 by 35 degrees. The online catalogue, which is almost two days behind, has been available to the public since June of 2002. The EQ selection windows are set for magnitude M (larger than Mc) and focal depth only for some small regions (less than 300 km). The value of Mc is dependent of the regional seismicity. For example, Mc is about 3.3-3.5 for small region and Mc is 4 for the large region. We then sequence the source parameters of every EQ in the chronological event order for each region. The EQ source parameters are the epicenterfs latitude (LAT) and longitude (LON), focal depth (DEP), inter-EQ time interval (INT) and magnitude (MAG). The findings of our experimental forecasting, started on June 2003 (www.tec21.jp), suggest that the following deterministic forecasting of large EQ (larger than about 6) for small regions is possible as forecasting of typhoons [Takeda, 2003; Takeda & Takeo, 2004]. 1) In months and weeks ahead of time. 2) The date of rupture within a day or two. 3) The hypocenter within about 0.5 degrees in LAT and LON and about 20km in DEP. 4) The magnitude within 0.5. Our physical model of the seismogenesis bases the forecasting. For example, the model can accurately predict the fault width of the large EQ. The model is an extension of so-called brittle-ductile interaction hypothesis envisioned by Aki [2003, 2004]. Furthermore, running sum of INT over some consecutive events shows the temporal state of stress accumulation in the region. This is because chaos analysis of the series finds that the running sum has only three independent dynamical variables, suggesting that they are three principal stress components in the region. Monitoring only the temporal variations of the stress accumulation in the large region is extremely useful in forecasting only the rupture date of major EQfs whose M is larger than about 7 within a day or two day accuracy. Thus, one can extend our time series analyses globally as long as any EQ catalogue similar to that of JMA is available. http://www.tec21.jp
S32B-05
Aseismic Slips Preceding Ruptures Assumed for Anomalous Seismicities and Crustal Deformations
If aseismic slips occurs on a fault or its deeper extension, both seismicity and geodetic records around the source should be affected. Such anomalies are revealed to have occurred during the last several years leading up to the October 2004 Chuetsu Earthquake of M6.8, the March 2007 Noto Peninsula Earthquake of M6.9, and the July 2007 Chuetsu-Oki Earthquake of M6.8, which occurred successively in the near-field, central Japan. Seismic zones of negative and positive increments of the Coulomb failure stress, assuming such slips, show seismic quiescence and activation, respectively, relative to the predicted rate by the ETAS model. These are further supported by transient crustal movement around the source preceding the rupture. Namely, time series of the baseline distance records between a numbers of the permanent GPS stations deviated from the predicted trend, with the trend of different slope that is basically consistent with the horizontal displacements of the stations due to the assumed slips. References Ogata, Y. (2007) Seismicity and geodetic anomalies in a wide area preceding the Niigata-Ken-Chuetsu Earthquake of October 23, 2004, central Japan, J. Geophys. Res. 112, in press. http://www.ism.ac.jp/~ogata/Ssg/ssgE.html
S32B-06
Testing the Daily Predictive Power of Clustered Seismicity Models on the 1992 Landers Aftershock Sequence
Aftershock hazard is a significant and strongly time-dependent contribution to seismic hazard. Large aftershock sequences can serve as natural laboratories for earthquake predictability experiments, allowing for retro- and prospective testing of seismicity forecasts on various spatial and temporal scales. The 1992 Landers earthquake and the resulting highly clustered earthquake sequence is one of the best recorded and best studied earthquake sequence; however, a comparative experiment to retrospectively forecast earthquakes applying established statistical and physical models has not yet been pursued. We analyze the performance of (1) the Short-Term Earthquake Probability (STEP) model and STEP model elements, (2) a suite of Epidemic Type Aftershock Sequence (ETAS) models with and without parameter dependence on time and space and various spatial triggering kernels, and (3) a model deriving seismicity rates from a rate and state model incorporating multiple stress changes due to large and moderate earthquakes in the aftershock sequence. Comparing these models allows us to address the questions: Which models perform well on short time and small spatial scales? Do physical models lead to an information gain over purely statistical models on the scale of an aftershock sequence? Where with respect to the faulting do the models perform well? For the experiment, we define rules similar to the RELM testing approach. Forecasts are computed for 90 days starting on June 28,.1992, forecasting the seismicity on a predefined grid in the magnitude range 4 ≤ M ≤ 8 each day for 24 hours. The forecasts are evaluated on a daily basis using the RELM likelihood tests that test data consistency and relative performance of the models. Preliminary results show that the statistical models perform well in the long run of the earthquake sequence but not at the onset of the sequence, due to the lack of sequence specific information. We plan to investigate additional well recorded aftershocks sequences from a range of tectonic settings, such as the Colfiorito earthquake sequence in Italy or earthquake sequences in Iceland. The ultimate goal of the retrospective experiments in the framework of the EU project Seismic Early Warning For EuRope (SAFER) is to improve the ability of the models to forecast in real-time short-term earthquake probabilities for earthquake sequences.
S32B-07
RESEARCH OF HYDRO-GEOLOGICAL PRECURSORS OF EARTHQUAKES IN ARMENIA
The observations of hydro-geological regime of underground waters in observed boreholes began in Armenia in 1986. Now these work is concentrated in National Seismic Service. For a long time observations are carried out studying several parameters (debit, temperature, chemical and gas composition) in several deposits of carbon mineral waters of Armenia. The interpretation of materials shows that that a number of strong and medium-strength earthquakes are accompanied by anomal changes in the level of underground waters. Regarding mineral waters, in connection with earthquakes some parameters are immediately changed: debit, temperature, chemical and gas composition. The study of hydrogeodynamic characteristics of precursors specify that the quantity of registered hydrogeodynamic precursors decreases with the increase of epicentrical distance. The majority of precursors is registered at the distance of 200 km from epicenter. There is a tendency of gradual increase of time and amplitude of a precursor of an earthquake depending on the rise of magnitude and epicentral distance. The behaviour of hydrogeodynamic precursors depends on the angle between the faults, to which this or that borehole reaches; with increase of this angle the deformation in the zone of the fault during the preparation of earthquakes is stronger, than in terms of small angles. 1. S1 2. Earthquake processes, Precursors and Forecasts 3. Garni Geophysical Observatory of the National Academy of Sciences of Armenia, 375019, Yerevan, Republic of Armenia, email: hakhleon@sci.am 4. O 5. 10808801 6. Artavazd Payment Type: select 'Purchase Order' PO Number: AGU WAIVER Billing Address: Enter Your Institution City: Enter Your City Country Code: Enter Your Country Name: Enter Your Name Phone: Enter Your Telephone Number
S32B-08
Methods of Multivariable Earthquake Precursor Analysis and a Proposed Prototype Earthquake Early Warning System
Significant advances are being made in earthquake prediction theory; however, a reliable method for forecasting the occurrence of earthquakes from space and/or ground based technologies remains limited to no more than a few minutes before the event happens. Several claims of earthquake precursors have been put forward, such as ionospheric changes, electromagnetic effects, and ground heating, though the science behind these is far from complete and the successful application of these precursors is highly regionally variable. Existing and planned dedicated space missions for monitoring earthquake precursors are insufficient for resolving the precursor issue. Their performance does not satisfy the requirements of an earthquake early warning system in terms of spatial and temporal coverage (Pulinets and Boyarchuk, 2004). To achieve statistically significant validation of precursors for early warning delivery, precursor data must be obtained from simultaneous repeated monitoring of several precursors in focus regions over a long period of time and then integrated and processed. Data sources include historical data, data from ground-based units, airborne systems, and space-based systems. This paper describes methods of systematic evaluation of regionally specific, multivariable precursor data needed for the identification of the expected time, magnitude and the position of the epicentre. This data set forms the basis for a proposed operational early warning system developed at the International Space University and which is built in partnership with local and national governments as well as international organizations.