Figure S2. Magnitude calibration, individual earthquakes are shown as black dots. (a) The abnormal amplitudes at around M1 result in biased linear fitting (slope=0.78). (b) After removing M0.83-M1.40 earthquakes (gray dots) in the calibration, we obtain slope=0.92 and the linear fitting is more reasonable. Note that the earthquakes removed from calibration are still included in the spectral analysis.