Estimating interpretation uncertainty from magnetotelluric inversion

Summary

We present a depth to basement case study which incorporates knowledge of magnetotelluric inversion uncertainty into the final interpretation.

Inversion with Occam2D and rj-McMC was conducted for over 1600 magnetotelluric sites from the Isa Extension MT survey in western Queensland. The combination of 2D deterministic inversion and 1D probabilistic inversion offered surety that the 2D data were being modelled adequately, and provided a means of estimating inversion uncertainty. Synthetic modelling was also used to establish the resolvability of the basement interface based on the resistive character of the project area; derived from downhole resistivity data.

The final depth to basement surface was interpreted jointly from the 1D and 2D inversion results. Two sources of uncertainty were independently estimated for the interpretation. (1) Uncertainty due to inversion nonuniqueness was estimated by assessing the scatter in 1D inversion results. (2) Maximum and minimum possible basement depths for the final interpretation were estimated from the probability distributions from 1D inversion combined with knowledge of basement resolvability from synthetic modelling.

Our study provides a new depth to basement interpretation with robustly determined error margins. The final interpretation is paired with maximum and minimum bounds to reflect the poorly constrained nature of the basement interface. Comparison to drilling results suggests the error estimates are appropriate. The study demonstrates the limitations of using magnetotelluric data to model depth to basement in the absence of independent constraining information.

• magnetotelluric
• inversion uncertainty
• interpretation