Abstract:

Time-lapse monitoring is an effective way to observe underground dynamic changes. Geophysical inversion results could be affected by systemic factors, including measurement errors, noise pollution in the studied environment, and the inversion ambiguity. If time-lapse geophysical data is inverted separately at different times, the results may have poor comparability, which would affect the monitoring of time-lapse geophysics. For the actual needs of the controlled-source audio frequency magnetotelluric (CSAMT) in the monitoring field, we integrated the observation data at different moments for inversion, and constructed resistivity models at different moments (mutually constrained) for three-dimensional inversion of the time-lapse CSAMT. Three types of models were designed to perform without time-lapse constraint inversion and time-lapse inversion comparison trial calculations of synthetic data, which verified the inversion algorithm and its effectiveness. The inversion results show that by adding adjacent time model constraints, the time-lapse controllable source inversion can better focus on the anomaly locations, reduce the noise impact and local environment changes, and minimize the impact of inversion artifacts. Our findings have laid a good foundation for the application of time-lapse CSAMT.

Key words: controlled-source audio-frequency magnetotellurics, limited-memory quasi-Newton method, time-lapse constraint