A new procedure for locating local earthquakes is proposed.
Essentially, this procedure consists in solving—by means of least-squares technique—a system of equations which is formally analogous to that of Geiger (Ax = b) but different from his in the values of the matrix A and vector b elements. This difference makes our procedure more reliable than Geiger's because it significantly reduces the cases of both iterative process divergence and low precision. Among the factors contributing to this progress the lesser possibility that the matrix A contains columns proportional (or nearly proportional) one to another is considered of particular influence.
More than 20,000 hypocentral calculations have been performed on simulated shocks: significant differences in the number of good locations were revealed between our procedure and the classical method of Geiger (1910). Ours is more precise, particularly when few stations were used or networks with an unsatisfactory geometry. The earth model also influences the observed differences as it contributes to generating those analytical conditions which make the calculation convergence more problematic when applying Geiger's method.
Further applications are currently carried out in order to verify the procedure features for velocity laws and station configuration different from those used in this study.