Estimating Noise Levels in AEM Data

Abstract

This paper reports the results of analysing AEM data where the same flight line has been flown repeatedly to monitor system performance. When these data are corrected for the effects of variable survey altitude they provide a good measure of the reproducibility of AEM data. The analysis has been conducted for both frequency and time domain systems and shows that, where the area is even moderately conductive, multiplicative errors will provide the dominant source of noise for AEM surveys. Errors typically have standard deviations of 2 % and can easily induce fluctuations of 10%. Moreover, because of their geometric origins, these errors are highly correlated between channels. Frequency domain HEM and Z component time domain systems produce errors that are largely unrelated to time/frequency/conductivity. This is not the case for X component data from asymmetric systems.