11:00 AM - 1:00 PM
[STT38-P07] Simple hypocenter determination method using the back-propagating wavefront from receivers
Keywords:hypocenter determination method
When we determine the hypocenter location, grouping observed travel time data according to each event and discrimination of P phase and S phase are important processes. It tends to be difficult to carry out the processes when many events occur. We develop a new simple method not needing such processes. The method uses only the time information of the observed travel time data at each seismic station. Rolling back the time, both of P phase and S phase waves start to propagate at either of P and S picked times on the stations. The location where many waves overlap is an event location. The wave is isochrone surface of travel time calculated by ray tracing method. All the travel time data of each station are already calculated and called when needed. A point is given to the grid point of space and time where the wave front exists. The grid point of high score over a given threshold is adopted as an event location and the origin time. Target area and time are discretized. The precision of the location and the time depend on the discretized grid size. Too small size requires too long calculation time and too large computer memory.
To check this method works fine, we assume 100km-long cubic area, the events, and the stations and give the calculated travel time data at each station. Using the travel time data, we confirm the assumed events are successfully inferred by this method. Grid intervals in space and time are 0.5km and 0,1s. The seismic velocity of P(Vp) is uniformly 10km/s and Vs = Vp/1.7 in all target area. When 8 events occur at the same time, they can be determined separately if there are 16 stations. We find that irregular station distribution (as in real seismic network) is better than regular one. When the stations are regularly located at each grid point, several high score areas appear not at the assumed event location. We confirmed our method works fine in polar coordinate system, too. The computation time is short enough (one order or much smaller than the analyzed data time window).
Our study shows that it will be possible to grasp an image of event distribution with our method even if many events occur.
Grouping the picked travel time data by event and classifying them according to the phases will be possible if we compare the calculated travel times of the relocated event by our method with the picked ones. The grouped and classified picking data are available to precisely determine the event location by the ordinary event relocation method.
We will check our method using real events, real stations and real travel time data.
To check this method works fine, we assume 100km-long cubic area, the events, and the stations and give the calculated travel time data at each station. Using the travel time data, we confirm the assumed events are successfully inferred by this method. Grid intervals in space and time are 0.5km and 0,1s. The seismic velocity of P(Vp) is uniformly 10km/s and Vs = Vp/1.7 in all target area. When 8 events occur at the same time, they can be determined separately if there are 16 stations. We find that irregular station distribution (as in real seismic network) is better than regular one. When the stations are regularly located at each grid point, several high score areas appear not at the assumed event location. We confirmed our method works fine in polar coordinate system, too. The computation time is short enough (one order or much smaller than the analyzed data time window).
Our study shows that it will be possible to grasp an image of event distribution with our method even if many events occur.
Grouping the picked travel time data by event and classifying them according to the phases will be possible if we compare the calculated travel times of the relocated event by our method with the picked ones. The grouped and classified picking data are available to precisely determine the event location by the ordinary event relocation method.
We will check our method using real events, real stations and real travel time data.