Improvement of near-surface resolution of airborne time-domain data. Optimizing inversion settings and applying system response in the inversion of real AEM data and synthetic data.

Airborne electromagnetics (AEM) is a geophysical investigation method used to make resistivity models of the subsurface, for ground investigations related to various types of projects. It was originally developed as an exploration method for use in the mining industry, but the development of the method happening around year 2000 opened up for new applications. More recently, AEM has been increasingly applied in geotechnical and engineering projects, demanding high-resolution information. The more recent use of AEM stretches the methods ability to give detailed subsurface information, especially in the near-surface. The first five to ten meters have the highest impact on surface infrastructure projects and are often the most difficult to resolve. Manufacturers and programmers continuously work to develop and improve the operational systems and the programs used to handle the acquired data, in order to achieve more accurate data and high-resolution models. Data from AEM surveys carried out in Norway, to support ground investigations for infrastructure projects, are used in this study. In large infrastructure projects, knowledge of sediment type and thickness is vital, as is information about possible occurrence of highly sensitive clay. Both a newly introduced method, utilizing earlier time data, is tested and experimentation with inversion settings is conducted, in order to investigate whether any improvement in the near-surface resolution can be gained for two datasets formerly acquired at NGI. In an area with conductive clay over resistive bedrock, the recently introduced system response method is tested. It is applied during the inversion of SkyTEM data and makes it possible to utilize the very earliest time gates (before 10 µs). By including earlier time data more information about the upper meters is obtained, which ideally should give an increase in the near-surface resolution of the models. The method is tested on a site where very small resistivity contrasts (5 to 10 Ωm embedded in 10 to 50 Ωm) are crucial to resolve to identify hazardous quick clay. The models obtained show to give more pronounced structures in the near-surface, reflecting true structures observed in resistivity borehole measurements. The same outcome is observed when conducting synthetic modelling. Utilizing the very early time gates thus increases the near-surface resolution and makes it possible to distinguish and resolve layers with small resistivity contrasts. In another setting, the stability and resolution of AEM models from a site with extremely high resistivity contrast (100s to 1000s Ωm overlaying 0.1 to 1 Ωm) in combination with a noisy environment is studied. Preliminary AEM models had a tendency to overestimate the thickness of the resistive overburden. The acquired dataset is reprocessed and inverted with different inversion settings, to investigate the effect the parameters set in the inversion have on the accuracy and resolution of the resulting models. Experimenting with and optimizing the inversion settings results in models better fitting other prior information from the survey area. This shows the importance of customizing the inversion settings to suit the geological condition present, and not just use default settings adapted to a more general case. Especially for unusual geological conditions, the inversion parameters should be carefully selected in order to obtain reliable models. In addition, limited low moment data were available due to a noisy environment. This affects the reliability of the models, illustrated by modelling and seen in the resulting real models.