Development of a machine learning model for seismic hazard from wastewater injection

The Railroad Commission of Texas is the state agency with primary regulatory jurisdiction over the oil and natural gas industry. For this project, ESMIA worked with the business unit of the Railroad Commission of Texas to complete a thorough review and re-design of an existing seismic machine learning model to improve understanding of risk factors for wastewater injection-induced earthquakes. In Phase I, ESMIA provided a recommendation report containing iterative improvements based on academic literature review and industry best practices using progressive learning techniques that may potentially yield greater accuracy. In Phase II, ESMIA focused on three key items: developing the input data model for improved feature quality and increased dataset size, implementing a multi-step machine learning approach to evaluate hazard using an adaptable spatial grid and time windows to improve robustness, and providing technical documentation and training on the new model for client-side integration.