[S19P-01] An attempt to estimate an appropriate distance correction function for local magnitude calculation for the Philippines
We analyzed broadband waveform data from the Philippine Seismic Network to estimate a distance correction function for local magnitude calculation appropriate for shallow events in and around the Philippines. We used data from the shallow (up to 35 km) events that occurred between July 2012 and 2022 whose magnitudes are equal to or greater than Mw 4.5.
First, we calculated local magnitudes using the distance correction function for southern California (Hutton and Boore, 1987) with station correction obtained by the differential scheme (Uhrhammer et al., 1996). The magnitude residuals become smaller with respect to hypocentral distance at shorter distances and they become larger at longer distances for both horizontal and vertical records. This indicates the higher attenuation at the shorter hypocentral distances and the lower attenuation at the longer hypocentral distances. These changes occurred about 600-700 km and 500-600 km for the horizontal and vertical records, respectively. The amplitude data are mainly from S waves and their coda in the shorter distance range, while they are mainly from the time series from P arrivals to S arrivals in the longer distance range. This is likely to cause the observed changes of the magnitude residuals with respect to hypocentral distance.
Using horizontal component data recorded at hypocentral distances up to 600 km, we performed a grid search to estimate a single distance correction function with station correction estimation. We evaluated the grid search results by the root mean square of the magnitude residuals and their hypocentral distance dependence. The latter was represented by the maximum value among the averages of the magnitude residuals for each 100 km bin of hypocentral distance. The selected distance correction function is 1.1 log(r/100)+ 0.00285 (r-100)+ 3, where r is the hypocentral distance. We compared the local magnitudes to the moment magnitudes to find their correlations. The next step in estimating the distance correction function will be to expand the data set to include smaller events.
First, we calculated local magnitudes using the distance correction function for southern California (Hutton and Boore, 1987) with station correction obtained by the differential scheme (Uhrhammer et al., 1996). The magnitude residuals become smaller with respect to hypocentral distance at shorter distances and they become larger at longer distances for both horizontal and vertical records. This indicates the higher attenuation at the shorter hypocentral distances and the lower attenuation at the longer hypocentral distances. These changes occurred about 600-700 km and 500-600 km for the horizontal and vertical records, respectively. The amplitude data are mainly from S waves and their coda in the shorter distance range, while they are mainly from the time series from P arrivals to S arrivals in the longer distance range. This is likely to cause the observed changes of the magnitude residuals with respect to hypocentral distance.
Using horizontal component data recorded at hypocentral distances up to 600 km, we performed a grid search to estimate a single distance correction function with station correction estimation. We evaluated the grid search results by the root mean square of the magnitude residuals and their hypocentral distance dependence. The latter was represented by the maximum value among the averages of the magnitude residuals for each 100 km bin of hypocentral distance. The selected distance correction function is 1.1 log(r/100)+ 0.00285 (r-100)+ 3, where r is the hypocentral distance. We compared the local magnitudes to the moment magnitudes to find their correlations. The next step in estimating the distance correction function will be to expand the data set to include smaller events.