Relocation of major earthquakes in India from 1905 to 2011 using a non-linear probabilistic approach

We perform probabilistic hypocentral location inversion for 10 major earthquakes in India that occurred since the dawn of the seismological instrumental era in 1904. A non-linear inversion approach with an efficient global sampling algorithm is used to obtain an estimate of the probability density function (PDF) in 3D space for the hypocenter estimation. The objective function that is minimized is an equal-differential traveltime formulation, that quantifies the difference between the observed and synthetic traveltimes for a pair of stations. Significant improvement in hypocentral location is obtained for five earthquakes in our dataset that occurred before the installation of the WWSSN network. The improvement and standardization of the global seismometer network in the 1960s led to more accurate and consistent location estimates across multiple reporting agencies. For post-1960 events, our hypocenter estimates are consistent with those reported by other authors. In most cases our maximum likelihood location lies in the vicinity of a mapped active fault, thereby providing confidence in our solution. The improvement in location estimate is gauged by measuring traveltime residuals across all stations and the size of the error ellipsoid.