Contribution of travel time calculation errors to hypocenter uncertainty at the Groningen Gas Field

Seismic event location uncertainty estimates combine the effects of multiple sources of error. Common amongst these are errors in the picking of arrival times and differences between the true velocity structure and its model approximation. Here, we attempt to quantify systematic biases that arise from limitations of the travel time calculations themselves.

Fundamental to our approach is an assumption that our wavefield simulations deliver physically correct P- and S- wave travel time information for a perfectly defined 3D elastic model. If the elastic model is sufficiently representative of local structural complexity and physical properties, then deviations from the wavefield simulation arrivals and travel times calculated using other methods can be treated as method biases.

By quantifying the method biases in this way, different travel time methods can be compared and optimized with a degree of separation from epistemic unknowns. The hypocenter estimation technique applied will determine the consequences for absolute and relative location error. The resulting discrete systematic location biases and residual uncertainties can then be compared to traditional estimates of hypocenter uncertainty.

Implications for location error and uncertainty are illustrated using examples from the Groningen Gas Field in the Netherlands. However, analyses of this type could be applied anywhere that a reasonable representation of the 3D subsurface structure is available. In principle, the approach could be extended to assess the relative merits of network densification versus technology development for a given setting.