5:15 PM - 6:45 PM
[STT37-P06] Relative hypocenter determination of JMA unified earthquake catalogue by hypocenter determination method using multiple template earthquakes (hypoTD)
Keywords:hypoTD, hypoDD, relative hypocenter location
1. Introduction There exists significant heterogeneity in the seismic velocity structure of the Earth. This variation poses a challenge when determining the precise location of earthquake hypocenters. Waldhauser & Ellsworth (2000) introduced a powerful technique called hypoDD, which stands for Double-Difference hypocenter relocation. This method aims to improve the accuracy of hypocenter determination by accounting for velocity heterogeneity and is used by many researchers. However, hypoDD has the following limitations. It cannot be applied to earthquakes occurring over a wide area. Locating hypocenters for newly occurring earthquakes is difficult. We propose a method for determining the relative hypocenters using a large number of template earthquakes and relocate hypocenters of the JMA unified earthquake catalog.
2. Method Horiuchi et al. (2022) developed an automatic hypocenter determination system with the capability to correct automatically picked arrival time data by computing relative hypocenters with several ten nearby earthquakes. The system is described in details on the web site https://www.homeseismo.com/html/sotaijido.html. This system includes the ability to determine the relative hypocenters of individual earthquakes. Here, we modified this system so that it can be applied to relative hypocenter location using arrival time data released by the Japan Meteorological Agency (JMA). Below is a description of this method. First, it registers the hypocenter parameters and P wave and S wave arrival time data of large earthquakes that occurred in the past within the analysis area. In the following, this large earthquake will be referred to as a template earthquake. For each earthquake of interest, N template earthquakes occurring near its hypocenter are selected. The goal is to find solution of hypocenter location that minimize the root mean square of the travel time residuals in N sets of relative hypocenter location. In this method N=50 template earthquakes are used. We will refer to this approach as hypoTD.
3. Results NIED releases epicenter parameters and arrival time data for the JMA unified earthquake catalog with a 2-3 day delay. We used this data. We conducted one iteration to determine the relative hypocenter by putting hypocenter parameters calculated by hypoTD as the hypocenter of the template earthquake. The analysis period is approximately five years from January 2019 to 2024. The earthquakes registered as template earthquakes are 270,000 earthquakes that occur in and around the Japan Islands, and whose hypocenters were determined by using P and S wave arrival times more than 20 and 10, respectively. The number of earthquakes whose hypocenters have been re-determined is 1.27 million. Figure 1 shows a comparison of the epicenter distribution of earthquakes in Tokyo Bay based on hypoTD and that of JMA unified earthquake catalog. As is clear from the comparison of the earthquake source distributions, the hypoTD earthquake sources are concentrated, making it appear that a highly accurate hypocenter distribution has been obtained. A similar comparison with a unified hypocenter in off the coast of Chiba Prefecture, the area in the western Kanagawa and eastern Yamanashi Prefectures, and other areas showed that hypoTD can provide highly accurate hypocenter distribution. The average RMS value of the travel time residuals of relative hypocenter location for 1.27 million earthquakes was 0.11 seconds. This value is much smaller than that of conventional hypocenter determination, indicating that hypoTD removes the influence of the heterogeneity of velocity structure.
2. Method Horiuchi et al. (2022) developed an automatic hypocenter determination system with the capability to correct automatically picked arrival time data by computing relative hypocenters with several ten nearby earthquakes. The system is described in details on the web site https://www.homeseismo.com/html/sotaijido.html. This system includes the ability to determine the relative hypocenters of individual earthquakes. Here, we modified this system so that it can be applied to relative hypocenter location using arrival time data released by the Japan Meteorological Agency (JMA). Below is a description of this method. First, it registers the hypocenter parameters and P wave and S wave arrival time data of large earthquakes that occurred in the past within the analysis area. In the following, this large earthquake will be referred to as a template earthquake. For each earthquake of interest, N template earthquakes occurring near its hypocenter are selected. The goal is to find solution of hypocenter location that minimize the root mean square of the travel time residuals in N sets of relative hypocenter location. In this method N=50 template earthquakes are used. We will refer to this approach as hypoTD.
3. Results NIED releases epicenter parameters and arrival time data for the JMA unified earthquake catalog with a 2-3 day delay. We used this data. We conducted one iteration to determine the relative hypocenter by putting hypocenter parameters calculated by hypoTD as the hypocenter of the template earthquake. The analysis period is approximately five years from January 2019 to 2024. The earthquakes registered as template earthquakes are 270,000 earthquakes that occur in and around the Japan Islands, and whose hypocenters were determined by using P and S wave arrival times more than 20 and 10, respectively. The number of earthquakes whose hypocenters have been re-determined is 1.27 million. Figure 1 shows a comparison of the epicenter distribution of earthquakes in Tokyo Bay based on hypoTD and that of JMA unified earthquake catalog. As is clear from the comparison of the earthquake source distributions, the hypoTD earthquake sources are concentrated, making it appear that a highly accurate hypocenter distribution has been obtained. A similar comparison with a unified hypocenter in off the coast of Chiba Prefecture, the area in the western Kanagawa and eastern Yamanashi Prefectures, and other areas showed that hypoTD can provide highly accurate hypocenter distribution. The average RMS value of the travel time residuals of relative hypocenter location for 1.27 million earthquakes was 0.11 seconds. This value is much smaller than that of conventional hypocenter determination, indicating that hypoTD removes the influence of the heterogeneity of velocity structure.