Improving Aeromagnetic Surveying Capabilities of Uninhabited Aircraft Systems

Uninhabited aircraft systems (UAS) have grown in popularity for aeromagnetic surveying. While the technology has been demonstrated to be viable, studies have not addressed three areas. First, comparisons with traditional platforms over geologically interesting regions are limited. Second, demonstrations of advanced processing with UAS data are rare. And lastly, methods for magnetic compensation of UAS data is outstanding. This thesis addresses these three areas and provide approaches to evaluate UAS performance and improve data quality. A hexacopter UAS was used to fly an aeromagnetic survey over a property with prospective gold targets. The UAS data was found to be repeatable and consistent. Qualitative and quantitative comparison with data from traditional magnetic surveys revealed that the UAS data could delineate geological structures better than the helicopter data and more efficient to collect than ground data. Unconstrained and constrained magnetic inversion demonstrated that the quality of the data collected by the UAS was sufficient to model the structural framework of banded iron formations within the survey area. It highlighted that the potential gold ore zones are not directly associated with them, but rather with steeply dipping faults that transect the area. The exercise showed that, at the early stage of exploration, performing unconstrained inversion yielded a realistic and detailed model of the subsurface, opening the possibility of implementing magnetic inversion as a continuous process during targeted high-resolution surveying for mineral exploration. Magnetic compensation of noise from aircraft attitude variations is typically modelled by performing a least-squares fit to a 16-term model by bandpass filtering data from a high-altitude (3,000 m) figure-of-merit flight. Government and hardware limitations generally prevent UAS to fly at such altitudes (over 122 m AGL), so an alternative solution was developed that uses recurrent neural networks on survey data, without the need of an FOM. The algorithm was tested on data from a traditional fixed-wing airplane survey and data from UAS flying at 120 m and 50 m above ground level. Comparisons with established compensation methods showed that the proposed algorithm could become a practical alternative.