Fig. 4
From: Forecasting the magnitude of the largest expected earthquake
Bayesian predictive distribution. PB(ΔT ≤ t|Sn, mex), Eq. (2), are plotted as solid curves and the corresponding probability density functions pB(t|Mn, mex) are plotted as histograms, for the sequence initiated by the magnitude M6.5 Kumamoto foreshock (April 14, 2016). The results correspond to the same early training time interval from Ts = 0.05 to Te = 2.16 days with all the events above magnitude mc ≥ 3.3 and for different forecasting extreme magnitudes mex = 5.8, 6.3, and 6.5. The horizontal lines correspond to the probability levels of 5% (dashed line), 10% (dash-dotted line), and 20% (dash-double-dotted line). The intercepts of these lines with the cumulative distribution functions give the forecasting or interarrival time intervals (provided in the figure legend) to the next largest expected earthquake with the magnitudes larger than mex
