Comparison of RMS-acceleration and Brune Stress Drops for Crustal Earthquakes In Japan
BALTAY, A., Stanford University, Stanford, CA, ; HANKS, T., United States Geological Survey,, Menlo Park, CA, firstname.lastname@example.org; PRIETO, G., University de los Andes, Bogotá, Bogotá, Colombia, email@example.com; IDE, S., University of Tokyo, Tokyo, Japan, firstname.lastname@example.org; BEROZA, G., Stanford University, Stanford, CA, email@example.com
The relationship between earthquake stress drop and ground motion acceleration is central to seismic hazard analysis. We measure root-mean-square peak and RMS (root-mean-square) acceleration using KikNet accelerometer data from Japan for three crustal earthquake sequences. From RMS acceleration measurements, we calculate the aRMS-stress drop, using the relationship of McGuire and Hanks . For these same earthquakes, we have used the empirical Green’s function – coda method of Baltay et al.  to estimate corner frequency and Brune stress drop. The aRMS stress drop compares well to the Brune stress drop, especially at stations close to the source. Stress drop based on aRMS is proportional to the inverse square root of the corner frequency, while Brune stress drop depends on the corner frequency cubed, so we expect less variation in aRMS-stress drops if the corner frequency uncertainty is the dominant source of uncertainty. We find, however, that the eGf-coda method estimates a slightly tighter log-normal distribution than the aRMS-stress drop method. The log-mean and log-std of stress drop is 5.1 MPa and 0.42 for the eGf-coda method, and 7.6 MPa and of 0.51 for the aRMS method. While our eGf-coda method for determining stress drop and other source parameters is more robust, the aRMS method can be used as a quick easy way to estimate the stress drop. RMS acceleration measurements correlate well with maximum acceleration, or PGA, as expected from random vibration theory [Vanmarcke and Lai, 1977]. This implies that RMS-acceleration and stress drop can quickly and easily be estimated only from the peak acceleration of a record. Over all, we find that earthquakes in Honshu, Japan, follow expected scaling laws and support self-similarity, and both the methods add extra validation to the omega -2 source model.