GEOTHERMAL SEISMOLOGY: THE STATE OF THE ART

Injecting fluid into crustal rocks for purposes such as engineering geothermal systems and sequestering CO 2 often has, as a side effect, the stimulation of seismic activity. Understanding the physical processes involved is important for controlling the maximum size of such earthquakes, and limiting the damage that they might cause. At the same time, the seismic waves from induced earthquakes provide a rich source of potentially high-resolution information about these physical processes. Many recent seismological developments, such as the momenttensor source representation, high-resolution relative hypocenter-determination, and time-dependent seismic tomography, have greatly advanced our ability to extract this information from seismograms, particularly when different result types (e.g. moment tensors and hypocenter locations) are interpreted jointly. Many challenges still remain, however. Confidence assessments for derived quantities are essential components of any scientific investigation. Methods for computing confidence regions for moment-tensor source mechanisms have not been available until recently, and are still a rapidly developing subject, especially for very small (micro-) earthquakes. Most common hypocenter-location computer programs use methods that involve highly unrealistic assumptions about the sources of errors, e.g., that the crustal velocity structure is perfectly known, and produce confidence regions that are too optimistic by an order of magnitude. In truth, hypocenter location errors are dominated by real geophysical travel-time anomalies, not seismogramreading errors. Methods based on stochastic modeling of wave-speed variations in the Earth can greatly improve both estimated hypocenters and estimated hypocentral confidence regions.