A decade-long between-system comparison of AEM conductivities at Menindee Lake using deterministic and stochastic inversion.

Airborne electromagnetic (AEM) surveying provides a rapid means of imaging shallow subsurface geology as represented by changes in electrical conductivity within the earth. Aircraft-borne systems fly at different heights and with different speeds, and the exciting transients for time-domain AEM systems provide different spectral content to image the earth with. Geoscience Australia operates a test range over one of the Menindee lakes, in New South Wales, Australia, where different AEM systems have been flown over nearly a decade. Due to well studied geology and downhole induction data available in the area, this test range provides a useful proving ground for new AEM technology. For every test survey, certain lines within the range are repeatedly flown, and high-altitude lines are also acquired, such that robust data noise statistics can be established for all overflying AEM systems. Test-range data and noise for various systems naturally allows us to compare AEM derived subsurface images of the test line. This study presents the results of both deterministic as well as Bayesian stochastic inversion over the same 13 km stretch of land, with six different systems flown between 2014-2023. While a deterministic inversion provides a first-pass image for comparing AEM systems, far more information is provided by the full posterior distribution of inverted conductivities, and in particular, the marginal quantiles of median and extremal conductivities over the entire image section. Our findings indicate that there is generally good agreement with borehole logs, and the posterior conductivities for all systems agree well at the regional scale. The uncertainty (or the lack thereof) around ambiguous features in deterministic inversions is revealed through the stochastic inversions. Finally, we note that examination of water volumes in Menindee lakes do not show a simple relationship with inferred conductivity, indicating that unentangling environmental factors and system differences is a non-trivial matter.