1. Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ? Kunihiko Shimazaki (The Association for Earthquake Disaster Prevention ) Wahyu Triyoso （ ITB ） #Takeo Ishibe (ERI. Univ. of Tokyo)

2. Niigata-Kobe tectonic zone from short-term GPS observation （ Sagiya et al., 2000) Is this a permanent feature? We examine it from long-term historical seismicity data. Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ? Location of historical large earthquakes ⇔ Earthquake Occurrence rate estimated from GPS data

3. Relationship between strain rate and moment release rate Earthquake occurrence rate can be estimated from moment release rate, b-value and the G-R relationship Flowchart of quantitative comparison of location of historical large earthquakes with strain rate observed by GPS network

4. GPS dataStrain rate Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ?

5. Deformation rate is estimated from GEONET data (1994-1999) by Geographical Survey Institute (GSI)

6. Spatially smoothed (red) by using Elfiky and Kato (1999)

7. GPS dataStrain rate Moment release rate (a-value) Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ? Relationship between Horizontal strain and earthquake moment does not uniquely determined; however, the result does not depend on relationships we used; we used equation of Ward (1994) Ward (1994) WG on Calif. Eq. Probabilities (1995) Savage and Simpson (1997)

8. GPS dataStrain rate Moment release rate (a-value) G-R relationship (b-value is assumed to be 0.9) Spatial distribution of earthquake occurrence rate Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ?

9. Earthquake occurrence probability can be estimated from moment release rate and the G-R relationship (b-value is assumed to be 0.9)

10. GPS dataStrain rate Moment release rate (a-value) Location of historical Large earthquakes G-R relationship (b-value is assumed to be 0.9) Spatial distribution of earthquake occurrence rate What does GPS data represent? Spatial Correlation between Strained Regions and Large Shallow Crustal Earthquakes in Japan: Which is the cause ? Consistent or inconsistent ?

11. Historical earthquakes data Basically based on Usami (2003). Aftershocks and Aki- nada, Iyo-nada earthquakes are eliminated Jan.1596-May 2007 （ Chuetsu-oki earthquake is not included) Magnitude is 6 ¾ or M6.8 and above (total number is 52). The target interval is divided into 5 intervals (1596-1789, 1670-1855, 1729-1914, 1828-1945, 1896-Mar 2007 ） each interval includes 20 earthquakes (12 earthquakes overlap)

12. 1596- 1670- 1729- 1828- 1896- 1789 1855 1914 1945 May 2007 Comparison the probability of earthquake occurrence with location of historical earthquakes

13. Quantitative discussion by using dAIC Strain release model estimated from GPS data vs. no information model (earthquake occurrence rate is uniform throughout Japan) Likelihood function Cell with no earthquakes cell with earthquakes Occurrence rate is adjusted from factor k, k is determined to maximize likelihood function

14. 1596- 1670- 1729- 1828- 1896- 1789 1855 1914 1945 May 2007 -4.7 -2.8 4.6 -1.1 -11.2 dAIC GPS model is inferior to no information model except for 1729-1914.

15. 1596- 1670- 1729- 1828- 1896- 1789 1855 1914 1945 May 2007 -4.7 -2.8 4.6 -1.1 -11.2 ΔAIC 0.11 0.12 0.12 0.18 0.19 k K-value is probability estimated from earthq. Occurrence rate/ GPS ~0.1-0.2 (including the effect of subduction of plate)

16. /y Deformation rate caused by subduction of plate is calculated based on Savage(1983). This rate is subtracted from deformation rate estimated from GPS data Removing the effect of subduction of plate

17. excluding the effect of subduction of oceanic plate Including the effect of subduction of oceanic plate

18. 1596- 1670- 1729- 1828- 1896- -1.9 0.1 4.2 1.8 -4.8 dAIC 0.17 0.19 0.19 0.29 0.31 k 1789 1855 1914 1945 May 2007 Historical large earthq. occurred during 1729-1914 are well correlated with recently observed GPS data (about 100 times more likely to occur compared to earthq. Occurred after 1896.)

19. Activity and distribution of late Quaternary active fault (left) and earthquake occurrence rate based on horizontal strain (right). HERP （ 2005 ）

20. GPS data reflects temporary phenomena Tectonic lines with high activity such as Itoigawa-Shizuoka tectonic line (ISTL), median tectonic line (MTL), can not be recognized from GPS data

21. Conclusion If present deformation rate is permanent, it is inconsistent with distribution and activity of late Quaternary active fault On the contrary, if we consider that present deformation is temporary, it is thought to be caused by historical earthquakes occurred 100-300 years ago (flux deformation in mantle and/or lower crust is triggered after large earthquakes.) Researches of long-term triggered flux deformation are important to detect precursor changes accurately.

22. Thatcher, Matsuda, Kato, and Rundle (1980) 地殻内地震の長期余効 変動の例（上下動） 1896 年 陸羽地震

23. 1596- 1670- 1729- 1828- 1896- -1.9 0.1 4.2 1.8 -4.8 ΔAIC 0.17 0.19 0.19 0.29 0.31 k 1789 1855 1914 1945 Mar 2007 値 k から、 7 割は非地震性の地殻変動が起こっている。 又は、 Savage(1983) では水平成分の説明が不十分。