Location and magnitudes of earthquakes in Central Asia from seismic intensity data: model calibration and validation

In this study, we estimate the location and magnitude of Central Asian earthquake from macroseismic intensity data. A set of 2373 intensity observations from 15 earthquakes is analysed to calibrate non-parametric models for the source and attenuation with distance, the distance being computed from the instrumental epicentres located according to the International Seismological Centre (ISC) catalogue. In a second step, the non-parametric source model is regressed against different magnitude values (e.g. M-LH, m(b), M-S, M-w) as listed in various instrumental catalogues. The reliability of the calibrated model is then assessed by applying the methodology to macroseismic intensity data from 29 validation earthquakes for which both M-LH and m(b) are available from the Central Asian Seismic Risk Initiative (CASRI) project and the ISC catalogue. An overall agreement is found for both the location and magnitude of these events, with the distribution of the differences between instrumental and intensity-based magnitudes having almost a zero mean, and standard deviations equal to 0.30 and 0.44 for m(b) and M-LH, respectively. The largest discrepancies are observed for the location of the 1985, M-LH = 7.0 southern Xinjiang earthquake, whose location is outside the area covered by the intensity assignments, and for the magnitude of the 1974, m(b) = 6.2 Markansu earthquake, which shows a difference in magnitude greater than one unit in terms of M-LH. Finally, the relationships calibrated for the non-parametric source model are applied to assign different magnitude-scale values to earthquakes that lack instrumental information. In particular, an intensity-based moment magnitude is assigned to all of the validation earthquakes.