A procedure is developed for objectively analyzing the geometrical configuration of a small seismographic station network. The elements of the parameter space covariance matrix are used to determine the uncertainty in the hypocentral parameters and their corresponding cross-correlations for a given network geometry. Likewise, the elements of the data space covariance matrix are used to determine which data supply the least information toward the solution of the system of equations and which combinations of observations are redundant. Knowledge of the variation of these covariance matrices in the vicinity of a network allows an objective basis for comparing the performance of proposed network geometrical configurations.
A small triangular quadripartite network (10-km radius) is analyzed as an example of the method. For small networks, generally the most important data are S-wave onset times at the peripheral stations and P-wave onset times at the interior stations. The single most useful indicator of the performance of a proposed network configuration is the span of the semi-major axes of the covariance matrices for hypocenters in the vicinity of the network.