Because of the number of free parameters in a multiple-event relocation analysis, as compared to a single event relocation, mloc is sensitive to the accuracy of those initial hypocenters. In other words, if the initial locations are very poor mloc may have trouble converging. Hypocenters for recent events obtained from the major global and regional seismograph networks are usually adequate, within 20 km or so of the ultimate location, but older events are prone to having much larger errors. In such cases several runs may be needed in which mloc does not converge because the relative location (cluster vector) of one or more events continues to shift. In the early stages of a relocation analysis where such issues are being addressed it is highly recommended to leave focal depth as a fixed parameter.
Sources of Initial Hypocentral Parameters
mloc can take the initial estimate of location from several sources, in order of decreasing precedence:
- A command (commands time, dep_, lat or long) in the event definition section of the command file.
- A value taken from the ~.hdf file of a previous run (command rhdf). Depending on the type of relocation this may be a file named ~.hdf, ~.hdf_cal or ~.hdf_dcal.
- A command (usually depc, but in principle, time, lat or long) in the first section of a command file (before any events are defined) that will therefore apply to all events unless specifically over-ridden.
- The preferred hypocenter of the input data file.
For the initial run there is usually no alternative to using the locations carried in the input data file. Normally I will switch to taking starting hypocentral parameters from the ~.hdf file of the last run (command rhdf) immediately after the first run.