A METHOD FOR MULTIDIMENSIONAL-ANALYSIS OF EARTHQUAKE FREQUENCY-MAGNITUDE DISTRIBUTION WITH AN APPLICATION TO THE VRANCEA REGION OF ROMANIA

A novel method is proposed for the rime-space analysis of the frequenc y-magnitude distribution of earthquakes (FMD). The FMD is calculated u sing a non-parametric technique, based on the nearest neighbourhood st atistics, and the FMD parameters, namely the occurrence rate at a refe rence magnitude and the b-value, are subsequently estimated by a param etric (log-linear) regression. The method's potential is examined by u sing a microearthquake catalogue of the intermediate depth (h > 60 km) seismicity in the Vrancea region (Romania) spanning 18 yr. For the ma gnitude intervals of M(L) 2.8-4.0 and 4.0-5.2, the calculated occurren ce rates exhibited different linear decays with increasing magnitude o n a log-linear scale. Quasi-linear correlations were found between the FMD parameters over selected time-space intervals. In several cases, the results showed that changes in the b-values were inversely related to changes in the occurrence rates of magnitudes at the limits of the first magnitude interval. The three large events (M(W) similar to 7) reported in the catalogue were preceded by time-space regions of relat ively low b-values lasting for a few years at the depth of these event s. Three episodes of distinct behaviour were identified: (i) b-values are inversely related to changes in the occurrence rates of magnitudes at the limits of the first study range (M(L) 2.8 and 4.0); (ii) a dec rease in the b-value occurs when an increase in the occurrence rate of relatively larger magnitude events (M(L) 4.0) is observed; (iii) the occurrence rate of larger events tends to saturate, whereas the rate o f small events tends to fluctuate. The results suggest that multi-dime nsional b-values and occurrence rates over selected magnitudes have th e potential to outline characteristic trends in the recorded seismicit y, and provide relative constraints for the modelling of the earthquak e cycle.