第72回統計地震学セミナー / The 72nd Statistical Seismology Seminar
- Date&Time
- 2019年1月22日(火)
/ 22 January, 2019 (Tue) 13:30 – 16:30
- Place
- 統計数理研究所 セミナー室7 (A504)
/ Seminar room7 (A504) @ The Institute of Statistical Mathematics
- Speaker1
- Prof. Xiaofei Chen
(Southern University of Science and Technology)
- Title
- Phase diagram of earthquakes and implications
- Speaker2
- Kazuyoshi Z Nanjo
(University of Shizuoka)
- Title
- An investigation into the relation between the occurrence of large earthquakes and time-dependent decrease in b value
- Abstract
- The Gutenberg-Richter frequency-magnitude distribution of earthquakes is well established in seismology. The b value, the slope of the relation between frequency and magnitude is typically 1, but it often shows variations around 1. The b value has shown a pronounced decrease over several years prior to large earthquakes around their hypocenters. Specific examples include the M9-class 2011 Tohoku and 2004 Sumatra earthquakes (e.g., Nanjo et al., 2012). However, it has remained uncertain whether there is the existence of tendency that large earthquakes occur, following the appearance of b-value decrease. To prove this existence, we are now trying to create a method to make and evaluate trial retrospective forecasts of large earthquakes (e.g., M8+ earthquakes from 1980 to 2017 on the worldwide basis, using the ANSS catalog), based on decreasing trend in b values. This is still ongoing research, so that, in this talk, we present the preliminary result. Based on it, we then discuss the possibility that a decrease in b values can be considered as a precursor to large earthquakes and an important indicator that has potential in terms of forecasting large earthquakes.
- Speaker3
- Yuchen Wang
(Earthquake Research Institute, the University of Tokyo)
- Title
- Tsunami Data Assimilation in Disaster Mitigation
- Abstract
-
Tsunami data assimilation has been proposed for tsunami early warning. It estimates the tsunami waveform by assimilating offshore observed data into a numerical simulation, without calculating initial sea surface height at the source. The optimum interpolation method is adopted in data assimilation. However, previous data assimilation method has a relatively high computational load, as it is necessary to run numerical simulations to obtain the tsunami wavefield.
In our research, we proposed a new tsunami data assimilation approach based on Green’s function to reduce the computation time for tsunami early warning. Green’s Function-based Tsunami Data Assimilation (GFTDA) forecasts the waveforms at Points of Interest (PoIs) by superposition of Green’s functions between observation stations and PoIs. Unlike the previous assimilation approach, GFTDA does not require the calculation of the tsunami wavefield for the whole region during the assimilation process, because the Green’s functions have been calculated in advance. The forecasted waveforms can be calculated by a simple matrix manipulation.
This approach greatly reduces the time cost for tsunami warning because it no longer needs to run the tsunami propagation model, as long as the Green’s functions are calculated in advance. By combining with Huygens-Fresnel Principle, this method could be applied to regions without a dense observation network. The applications to the 2012 Haida Gwaii earthquake, the 2004 off the Kii Peninsula earthquake and the 2009 Dusky Sound earthquake revealed that GFTDA helped achieve a more accurate and quicker tsunami early warning while saving the cost.
