1. Yan Y. Kagan Dept. Earth and Space Sciences, UCLA, Los Angeles, CA 90095-1567, ykagan@ucla.edu, http://scec.ess.ucla.edu/ykagan.htmlykagan@ucla.edu EARTHQUAKE PATTERNS IN DIVERSE TECTONIC ZONES OF THE GLOBE http://moho.ess.ucla.edu/~kagan/Evison.ppt

2. Stochastic models of earthquake occurrence and forecasting Long-term models for earthquake occurrence, optimization of smoothing procedure and its testing (Kagan and Jackson, 1994, 2000). Empirical branching models (Kagan, 1973a,b; Kagan and Knopoff, 1987; Ogata, 1988, 1998; Kagan, 2006). Physical branching models – propagation of earthquake fault is simulated (Kagan and Knopoff, 1981; Kagan, 1982).

3. Kagan, Y. Y., 1991. Likelihood analysis of earthquake catalogues, Geophys. J. Int., 106, 135-148. Earthquake intensity function Time influence function (Omori’s law) Space influence function

4. Likelihood analysis of earthquake catalogues Y. Y. Kagan Geophys. J. Int. (1991) 106, 135-148 CALNET -- Central California; PDE – global; Harvard – CMT global; DUDA – global M>7.0.

7. PARAMETER VALUES FOR VARIOUS SUBDIVISIONS OF CMT CATALOG, 1982--2007/03/31, Mw>=5.6 -------------------------------------------------------- All Subd. Orog. Inter. Fast Slow -------------------------------------------------------- 1. N 7720 5022 1004 279 584 831 2 Mmax 9.07 9.07 8.28 8.15 7.67 7.15 3. Inf/N 1.03 1.18 1.11 0.59 0.30 0.44 4. Ind/N 0.745 0.690 0.819 0.866 0.935 0.941 5. \mu 0.131 0.169 0.093 0.099 0.042 0.035 6. b 1.02 0.98 0.96 1.05 1.28 1.23 7. \delta 0.43 0.39 0.37 0.30 0.35 0.39 8. \theta 0.12 0.11 0.27 *0.1 0.27 *1.0 9. \sigma 0.30 0.29 0.15* 0.47 0.17 0.27 10. \eps_r 21.8 22.1 21.6 18.0 19.5 17.7 11. \eps_h 3.5 4.4 3.0* 5.1 3.0* 3.0* -------------------------------------------------------- \sigma – focal size for M4 EQ; \eps_r – horizontal error; \eps_h – vertical error

8. Sumatra M 9.1 earthquake, t_m = 1.2 days

9. Landers, 1992, M=7.3; t_m=3.7 hours

10. PARAMETER VALUES FOR VARIOUS SUBDIVISIONS OF CMT CATALOG, 1982--2007/03/31, Mw>=5.6; close aftershocks removed (1.3%) ---------------------------- All All/c ---------------------------- 1. N 7720 7471 2 Mmax 9.07 9.07 3. Inf/N 1.03 0.857 4. Ind/N 0.745 0.758 5. \mu 0.131 0.131 6. b 1.02 1.01 7. \delta 0.43 0.41 8. \theta 0.12 0.1* 9. \sigma 0.30 0.28 10. \eps_r 21.8 21.9 11. \eps_h 3.5 3.4 ---------------------------- \sigma – focal size for M4 EQ; \eps_r – horizontal error; \eps_h – vertical error

11. World seismicity: 1990 – 2000 (PDE)

12. PARAMETER VALUES FOR VARIOUS SUBDIVISIONS OF PDE CATALOG, 1968--2007/01/01, M>=5.0 -------------------------------------------------------- All Subd. Orog. Inter. Fast Slow -------------------------------------------------------- 1. N 45942 29980 7686 2191 3296 2789 2. Mmax 8.80 8.80 8.50 8.45 7.60 7.30 3. Inf/N 1.90 1.99 2.16 1.40 1.28 0.98 4. Ind/N 0.680 0.661 0.687 0.716 0.816 0.869 5. \mu 0.141 0.140 0.133 0.234 0.182 0.109 6. b 1.05 1.04 1.05 1.17 1.17 0.93 7. \delta 0.42 0.43 0.43 0.23 0.02 0.12 8. \theta 0.28 0.33 0.23 0.12 0.40 0.55 9. \sigma 0.38 0.37 0.31 0.24 0.15* 0.20 10. \eps_r 9.5 9.7 7.9 9.8 10.1 12.6 11. \eps_h 3.0* 4.5 4.3 3.0* 3.0* 3.0* -------------------------------------------------------- \sigma – focal size for M4 EQ; \eps_r – horizontal error; \eps_h – vertical error

13. PARAMETER VALUES FOR VARIOUS SUBDIVISIONS OF PDE CATALOG,1968--2007/01/01, M>=5.0; CLOSE AFTERSHOCKS REMOVED (5.5%) -------------------------------------------------------- All Subd. Orog. Inter. Fast Slow -------------------------------------------------------- 1. N 42925 27648 7205 2127 3217 2728 2. Mmax 8.80 8.80 8.50 8.45 7.60 7.30 3. Inf/N 1.25 1.24 1.50 1.17 1.04 0.76 4. Ind/N 0.695 0.674 0.709 0.736 0.831 0.881 5. \mu 0.171 0.177 0.146 0.241 0.185 0.121 6. b 1.04 1.03 1.04 1.17 1.17 0.93 7. \delta 0.38 0.37 0.41 0.18 0.0* 0.0* 8. \theta 0.11 0.13 0.10 0.1* 0.32 0.39 9. \sigma 0.37 0.35 0.32 0.15* 0.15* 0.15* 10. \eps_r 9.7 9.8 7.8 9.7 10.0 12.5 11. \eps_h 3.0* 4.8 4.7 3.0* 3.0* 3.0* -------------------------------------------------------- \sigma – focal size for M4 EQ; \eps_r – horizontal error; \eps_h – vertical error

14. Moment likelihood map

15. Moment Magnitude Distribution

16. Likelihood function iterations Red – Likelihood function; Blue – Ratio indepen./N; Green – branching coefficient (\mu)

17. Challenges We need to start by investigating how different branching models of seismicity approximate global earthquake catalogs. Global catalogs have an advantage of being considerably less inhomogeneous in time and space than local catalogs, and there are no spatial boundary effects which greatly complicate the analysis of local catalogs. Local seismicity is controlled by a few aftershock sequences of strong events. It is important, however, to analyze local catalogs as well to see that model parameter values are similar for worldwide and local catalogs. See http://bemlar.ism.ac.jp/wiki/index.php/Bird%27s_Zones

18. END Thank you

19. Abstract We extend existing branching models for earthquake occurrences by incorporating potentially important estimates of tectonic deformation and by allowing the parameters in the models to vary across different tectonic zones. In particular we use the following short list of tectonic categories (similar to Bird & Kagan [2004]): (4) Trenches (including Subduction zones & Oceanic Convergent Boundaries, & earthquakes in outer rise or overriding plate); (3) Fast-spreading ridges (oceanic crust, spreading rate > 40 mm/a; includes transforms); (2) Slow-spreading ridges (oceanic crust, spreading rate < 40 mm/a, includes transforms); (1) Active continent (including continental parts of all orogens plus designated continental plate boundaries of Bird [2003]); (0) Plate-interior (the rest of the Earth's surface). We use these models to develop earthquake forecasts (i.e., maps of expected rate of occurrences of earthquakes above a given threshold magnitude) in several different categories: long-term (e.g., 5+ years, emphasizing spontaneous mainshocks) based on tectonic deformation and long-term clustering as recorded in catalogs; and short-term (e.g., daily, with emphasis on triggered seismicity). Each forecast is global to permit relatively rapid and conclusive testing.

20. Abstract (cont.) These forecasts differ in the weights they put on tectonic deformation versus instrumental seismicity, but all forecasts take account of the large differences in seismicity between different classes of plate boundary or different tectonic zones. See preliminary results at http://bemlar.ism.ac.jp/wiki/index.php/Bird%27s_Zones. To test the long-term forecast efficiency numerically we calculate the concentration diagram. To make these diagrams we divide the region into small cells (0.5 by 0.5 degrees) and estimate the theoretical rate of earthquakes above the magnitude threshold for each cell; count the events that actually occurred in each cell; sort the cells in decreasing order of theoretical rate; and compute cumulative values of forecast and observed earthquake rates as shown for two western Pacific regions in http://scec.ess.ucla.edu/~ykagan/tests_index.html (see FORECAST TEST FOR 2004-2006:). Similar plots have been used in several of our papers [Helmstetter et al., SRL, 2007; Shen et al., SRL, 2007]. In effect, these diagrams are equivalent to error diagrams proposed by Molchan [1990, 2003], and Molchan & Kagan [1992], but in this case instead of the temporal axis we use the spatial area as the horizontal axis. To make this concentration diagram look like an error diagram we flipped the plot along the horizontal line 0.5, so that, for example, for the NW Pacific region our model predicts that 75% of seismicity would be concentrated in 10% of the area.

21. Long-term forecast: 1977-today Spatial smoothing kernel is optimized by using the first part of a catalog to forecast its second part. Kagan, Y. Y., and D. D. Jackson, 2000. Probabilistic forecasting of earthquakes, Geophys. J. Int., 143, 438-453.

22. Time history of long-term and hybrid (short-term plus 0.8 * long-term) forecast for a point at latitude 39.47 N., 143.54 E. northwest of Honshu Island, Japan. Blue line is the long- term forecast; red line is the hybrid forecast.

23. The table below and accompanying plots are calculated on 2007/ 4/19 at midnight Los Angeles time. The last earthquake with scalar seismic moment M>=10^17.7 Nm (Mw>=5.8) entered in the catalog occurred in the region 0.0 > LAT. > -60.0, -170.0 > LONG. > 110.0 on 2007/ 4/16 at latitude -57.89 and longitude 148.14, Mw = 6.42 ____________________________________________________________________ LONG-TERM FORECAST | SHORT-TERM Probability Focal mechanism | Probability Probability M>5.8 T-axis P-axis M>5.8 ratio eq/day*km^2 Pl Az Pl Az eq/day*km^2 Time- Longitude | | | Rotation Time- dependent/ | Latitude | | | angle dependent independent v v v v degree ……………………………………………………………………………………………………… 154.0 -7.5 1.49E-08 67 24 16 251 58.6 6.17E-11 4.13E-03 154.5 -7.5 2.02E-08 68 18 10 262 60.9 7.17E-11 3.54E-03 155.0 -7.5 2.60E-08 75 18 13 222 28.5 1.53E-09 5.88E-02 155.5 -7.5 3.51E-07 71 20 19 210 8.7 7.66E-07 2.2 156.0 -7.5 6.72E-08 76 20 14 216 20.0 8.49E-07 13. 156.5 -7.5 3.10E-08 76 28 13 231 41.4 5.12E-07 17. 157.0 -7.5 1.90E-08 1 327 16 236 44.7 1.37E-07 7.2 157.5 -7.5 8.92E-09 77 45 13 229 46.6 4.57E-09 0.51 158.0 -7.5 7.42E-09 79 60 11 228 45.3 9.06E-11 1.22E-02 158.5 -7.5 1.05E-08 49 147 4 52 54.2 1.66E-10 1.58E-02 159.0 -7.5 7.64E-09 47 147 4 242 58.8 1.07E-10 1.41E-02 ………………………………………………………………………………………………………

24. Short-term forecast uses Omori's law to extrapolate present seismicity. Forecast one day before the recent (2006/11/15) M8.3 Kuril Islands earthquake.

25. KURILE ISLANDS SEISMICITY 2005-PRESENT (2007/04/22) LATITUDE 40-50N, LONGITUDE 150-160E 1 2005 8 1 4 40 47.09 154.22 15.0 5.70 0.4 68 286 5 29 Thr 2 2005 10 15 10 6 46.85 154.33 46.0 6.17 1.8 64 289 1 22 Thr 3 2006 8 20 3 1 49.58 156.87 35.6 6.05 1.2 72 350 10 229 Thr 4 2006 9 28 1 36 46.45 153.66 12.0 6.00 1.0 71 278 18 122 Thr 5 2006 9 30 17 50 46.36 153.50 12.0 6.65 9.3 69 296 21 130 Thr 6 2006 9 30 17 56 46.19 153.35 16.4 6.03 1.1 72 291 18 122 Thr 7 2006 10 1 9 6 46.44 153.68 12.0 6.63 8.8 70 297 20 127 Thr 8 2006 10 13 13 47 46.15 153.73 12.0 5.90 0.7 70 287 19 129 Thr 9 2006 11 12 21 27 48.21 154.77 48.6 6.00 1.0 80 299 10 120 Thr 10 2006 11 15 11 15 46.75 154.32 13.4 8.35 3400.0 60 302 30 123 Thr 11 2006 11 16 6 20 46.41 154.68 12.0 6.00 1.0 4 314 82 190 Nor 12 2006 12 7 19 10 46.23 154.44 13.8 6.40 4.0 2 322 86 83 Nor 13 2006 12 26 15 19 48.18 155.20 38.0 6.00 1.0 77 340 11 133 Thr 14 2007 1 13 4 23 46.18 154.80 12.0 8.14 1650.0 10 150 67 264 Nor 15 2007 1 13 17 37 47.04 156.44 26.6 6.02 1.1 5 176 48 272 Nor 16 2007 4 9 10 18 48.21 155.17 33.0 5.83 0.6 73 318 16 124 Thr

26. Forecast one day after the recent (2006/11/15) M8.3 Kuril Islands earthquake.

27. Forecast one day before the recent (2007/01/13) M8.1 Kuril Islands earthquake.

28. Forecast one day after the recent (2007/01/13) M8.1 Kuril Islands earthquake.

29. Forecast one day before the recent (2007/4/1) M8.1 Solomon Islands earthquake.

30. Forecast one day after the recent (2007/4/1) M8.1 Solomon Islands earthquake

31. Long-term Forecast Efficiency Evaluation We simulate synthetic catalogs using smoothed seismicity map. Likelihood function for simulated catalogs and for real earthquakes in the time period of forecast is computed. If the `real earthquakes’ likelihood value is within 2.5— 97.5% of synthetic distribution, the forecast is considered successful. Kagan, Y. Y., and D. D. Jackson, 2000. Probabilistic forecasting of earthquakes, Geophys. J. Int., 143, 438-453.

32. Conclusions We present an earthquake forecast program which quantitatively predicts both long- and short-term earthquake probabilities. The program is numerically and rigorously testable. It is ready to be implemented as a technological solution for earthquake hazard forecasting and early warning.

33. END Thank you