1 Contributions of Prof. Tokuji Utsu to Statistical Seismology and Recent Developments Ogata, Yosihiko The Institute of Statistical Mathematics,Tokyo and Graduate University for Advanced Studies Utsu (1975) 2 Ogata et al. (1982,86) Seismicity rate = Trend + Clustering + Exogeneous effect deep Intermediate Shallow Shallow seismicity Intermediate + deep seismicity 3 Seismicity rate = trend + seasonality + cluster effect Ma Li & Vere-Jones (1997) SEASONALITY CLUSTERING 4 5 Matsumura (1986) 6 Magnitude Frequency: Utsu (1965) b-value estimation Aki (1965) MLE & Error assesment Utsu (1967) b-value test Utsu (1971, 1978) modified G-R Law Utsu (1978) h-value estimation h = E[(M-Mc)2] / E[M-Mc]2 7 Magnitude Frequency: o Bath Law (Richter, 1958) D1 := Mmain-M1 = 1.2 Utsu (1957) ~ D1 = 1.4 Median based on 90 Japanese Mmain= >6.5 Shallow earthquakes 8 Magnitude Frequency: Bath Law (Richter, 1958) D1=Mmain-M1 = 1.2 Utsu (1961, 1969) Magnitude difference o Mainshock Magnitude 9 Magnitude Frequency: Bath Law (Richter, 1958) D1=Mmain-M1 = 1.2 Utsu (1961, 1969) ~ D1 = 5.0 – 0.5Mmain for 6 = < Mmain= <8 ~ D1 = 2.0 for Mmain<6 Magnitude difference o Mainshock Magnitude 10 Aftershocks Utsu (1961) The Omori-Utsu formula for aftershock decay rate t : Elapsed time from the mainshock K,c,p : constant parameters 11 Utsu (1961, 1969) 12 1981 Nobi (M8) Aftershock freq. Data from Omori (1895) Mogi (1962) 13 Mogi (1967) 14 15 Mogi (1962) t > t0 = 1.0 day Utsu (1957) -p (t ) = Kt (t > t0) l 16 Mogi (1962) Utsu (1961) Utsu (1957) -p (t ) = Kt (t > t0) l 17 Mogi (1962) Utsu (1961) Utsu (1957) -p (t ) = Kt (t > t0) l Kagan & Knopoff Models (e.g., 1981, 1987) 18 Utsu (1962, BSSA) 1957 Aleutian 1958 Central Araska 1958 Southeastern Araska 19 Ogata (1983, J. Phys. Earth) Relative Quiescence in the Nobi aftershocks preceding the 1909 Anegawa earthquake of Ms7.0 1891 1909 20 21 Ogata & Shimazaki (1984, BSSA) Aftershocks of the1965 Rat Islands Earthquake of Mw8.7 l(s) ti = L(ti) 22 Utsu & Seki (1954) Utsu (1969) log S = 1.02M – 4.01 log S = M – 3.9 log L = 0.5M – 1.8 Utsu (1970) Tokachi-Oki earthquake May 16 1968 MJ=7.9 Aftershocks Nov. 1968 - Apr. 1970 …AABACBCBBBAA… B vs C&A A … - - + -- + - ++ - +++ -- … B Count runs C 23 24 cf., Reasenberg and Jones (1989) Utsu (1970) Standard aftershock activity: Occurrence rate of aftershock of Ms is during 1 < t < 100 days (M0>=5.5), where p=1.3, c=0.3 and b=0.85 are median estimates. The constant 1.83 is the best fit to 66 aftershock sequences in Japan during 1926-1968 25 Utsu (1970) Secondary Aftershocks 26 Omori-Utsu formula: (Ogata, 1986, 1988) t j is occurrence time of jth event; M j is magnitude of jth event; l0 is background rate; and parameters are (l0 , K , c, , p ). 27 Omori-Utsu formula: (Ogata, 1986, 1988) Kagan & Knopoff model (1987) n (t ) = –3/2 Kt , = 0, t= > 10 a+1.5Mj t < 10 a+1.5Mj 27 Omori-Utsu formula: (Ogata, 1986, 1988) Kagan & Knopoff model (1987) –3/2 (2/3)(M-Mc) . y(M) n (t ) = 10 Kt , = 0, t > tM = t < tM 28 29 30 31 32 1926 – 1995, M >= 5.0, depth < 100km 33 1926 – 1995, M >= 5.0, depth < 100km 34 35 36 37 38 39 40 Asperities Yamanaka & Kikuchi (2001) 41 42 43 44 LONGITUDE Cooler color shows quiescence relative to the HIST-ETAS model 45 Probability Forecasting Multiple Prediction Formula(Utsu,1977,78) 46 P0: Empirical occurrence probability of a large earthquake. Pm: Occurrence probability conditional on a precursory anomaly m; m = 1, 2, …, M, where probabilities are assumed mutually independent. Then, the occurrence probability based on all precursory anomalies is: Multiple Prediction Formula(Utsu,1977,78) 47 P0: Empirical occurrence probability of a large earthquake. Pm: Occurrence probability conditional on a precursory anomaly m; m = 1, 2, …, M, where probabilities are assumed mutually independent. Then, the occurrence probability based on all precursory anomalies is: Multiple Prediction Formula(Utsu,1977,78) 48 P0: Empirical occurrence probability of a large earthquake. Pm: Occurrence probability conditional on a precursory anomaly m; m = 1, 2, …, M, where probabilities are assumed mutually independent. Then, the occurrence probability based on all precursory anomalies is: Aki (1981) Multiple Prediction Formula(Utsu,1977,78) 49 P0: Empirical occurrence probability of a large earthquake. Pm: Occurrence probability conditional on a precursory anomaly m; m = 1, 2, …, M, where probabilities are assumed mutually independent. Then, the occurrence probability based on all precursory anomalies is: where Multiple Prediction Formula 50 F := { Ongoing events will be FORESHOCKS } logit Prob{ F | location, magnitude, time, space } = … Utsu (1978), Ogata, Utsu & Katsura (1995, 96, GJI ) Multiple Prediction Formula 51 F := { Ongoing events will be FORESHOCKS } logit Prob{ F | location, magnitude, time, space } = logit Prob{ F | location of the first event } +… Utsu (1978), Ogata, Utsu & Katsura (1995, 96, GJI ) 52 Multiple Prediction Formula F := { Ongoing events will be FORESHOCKS } logit Prob{ F | location, magnitude, time, space } = logit Prob{ F | location of the first event } + logit Prob{ F | magnitude sequential feature } +… Utsu(1978) Utsu (1978), Ogata, Utsu & Katsura (1995, 96, GJI ) Multiple Prediction Formula 53 F := { Ongoing events will be FORESHOCKS } logit Prob{ F | location, magnitude, time, space } = logit Prob{ F | location of the first event } + logit Prob{ F | magnitude sequential feature } + logit Prob{ F | temporal feature of a cluster } +… Utsu (1978), Ogata, Utsu & Katsura (1995, 96, GJI ) Multiple Prediction Formula 54 F := { Ongoing events will be FORESHOCKS } logit Prob{ F | location, magnitude, time, space } = logit Prob{ F | location of the first event } + logit Prob{ F | magnitude sequential feature } + logit Prob{ F | temporal feature of a cluster } + logit Prob{ F | spatial feature of a cluster } - 3 x logit Prob{ F } Utsu (1978), Ogata, Utsu & Katsura (1995, 96, GJI ) 55 56 57 TIMSAC84-SASE version 2 (Statistical Analysis of Series of Events) SASeis DOS version SASeis Windows Visual Basic SASeis 2006 with R graphical devices and Manuals 58 Thank you very much for listening