CODA METHOD AT HIGH FREQUENCIES: RETRIEVING SOURCE PARAMETERS OF SMALL (M1-M2) EARTHQUAKES

We calculate source parameters of small earthquakes from coda derived source spectrum ratios, extending the coda method to higher frequencies and smaller crustal volumes. To validate our high frequency results, we compare our source parameters estimates to the ones obtained with a direct wave study for the same set of earthquakes. We investigate earthquake source scaling relationships using local and regional good quality recordings of the M5 2002 Au Sable Forks, NY, mainshock and aftershocks sequence (M4-M1), and regional recordings of other moderate Eastern North America (ENA) earthquakes. We successfully retrieve spectral ratios, with a clear corner frequency and omega squared fall off, for the small (M2) Au Sable Forks aftershocks recorded locally (~3 km to 12 km epicentral distances) at a high sampling rate (200 sps). The local data have maximum coda duration of 15 s after the S wave onset (triggered recording). We investigate if the records are long enough, by first testing the method with Parkfield, CA, earthquakes, a dataset with similar dimensions in terms of station epicentral distances, earthquake magnitudes, and instrument sampling rate, for which records with long codas are available. The method holds for both datasets at high frequencies (up to 80 Hz), with a small increase of inter-station coda amplitude instability (0.5 standard deviation), expected for smaller earthquakes. We obtain in average higher corner frequencies and stress drops estimates (factor of 1.7 for corner frequency and 4.3 for stress drop) for the small (~M2) locally recorded Au Sable Forks earthquakes, than the values obtained with a direct wave study, and in average lower corner frequencies and stress drops for the moderate regionally recorded earthquakes (factor of 0.7 for corner frequency and 0.4 for stress drop). The corner frequencies of the mainshock and M3 earthquakes are close to the usable band limit and so uncertainties in the estimates of corner frequency and stress drop are high. We find high stress drops (median ~157 MPa) for this set of intraplate earthquakes and observe constant scaling of stress drop with earthquake size within the scatter and uncertainties. In the process of generating source spectra derived from regional coda envelopes, we empirically calibrate 8 regional broadband stations and obtain stable single-station seismic moment estimates for ENA in good agreement with local and Nuttli magnitude estimates form the regional networks that operate in ENA.