Spatio-Temporal Colocation Causality Analysis of Injected Water and Induced Earthquakes in Oklahoma

Abstract:

Water injection as a result of subsurface operations alters pore pressure that in turn increases the probability of human-made or induced earthquakes. Numerical models that establish relationships between water injection and induced earthquakes exist, however forecasting induced earthquakes as a result of water injection at field scale remains to be a challenge. A spatio-temporal causality analysis that leverage colocation statistic is proposed to model data-driven space-time relationships between water injection and earthquakes. Induced earthquake occurrence is modelled using space-time lags of water injection rate at injection wells via a colocation-based Grainger causality test. Earthquake data is obtained from USGS’ earthquake database and water injection is obtained from the public reports from Oklahoma Corporation Commission’s salt water disposal records. Results indicate spatial distribution of temporal lags between water injection and induced earthquake. Results show that impact of water injection in Oklahoma on induced earthquakes show high level of heterogeneity in terms of the time lag. Model developed for Oklahoma is further evaluated using experimental design to define safe water injection that avoids inducing earthquakes given previous occurrence of induced earthquakes.