Crustal permeability generated through microearthquakes is constrained by seismic moment.

Area of Science:

• Geophysics
• Earthquake Science
• Hydraulic Fracturing

Background:

• Understanding crustal permeability is crucial for subsurface energy applications.
• Microearthquakes (MEQs) can indicate changes in fluid flow and rock stress.

Purpose of the Study:

• To establish a quantitative link between microearthquake (MEQ) characteristics and crustal permeability evolution.
• To develop predictive models for permeability changes during hydraulic stimulation.

Main Methods:

• Conducted two field hydraulic stimulation experiments, simultaneously recording MEQs and permeability.
• Utilized a Bidirectional Long Short-Term Memory (Bi-LSTM) neural network model for permeability prediction.
• Applied transfer learning to test model generalizability across different experimental sites.

Main Results:

• The Bi-LSTM model accurately predicted permeability evolution and the extent of permeability increase.
• Identified key MEQ features that correlate with permeability changes, providing mechanistic insights.
• Demonstrated that transfer learning successfully predicted permeability evolution from an alternate dataset, confirming the inherent link between seismicity and permeability.
• Proposed a scaling relationship ($$ $$ ) between permeability changes and MEQ seismic moment for reactivated fractures.

Conclusions:

• MEQ features offer valuable insights into crustal permeability dynamics.
• The established scaling relationship provides a robust method for characterizing permeability evolution using MEQs.
• Findings support the use of MEQs as a tool for monitoring and predicting subsurface permeability changes.