Agglomerative hierarchical clustering of airborne electromagnetic data for multi-scale geological studies

Airborne electromagnetic methods provide detailed subsurface resistivity imaging over extensive areas. The inversion of electromagnetic measurements can be conducted with a quasi-3D spatially constrained inversion scheme, which yields numerous vertical resistivity soundings. So as to conduct the interpretation, these soundings can be interpolated to obtain a 3D resistivity model. However, large surveys result in huge resistivity models, which can be challenging to interpret with 2D or even 3D views. We propose a complementary approach for the interpretation of airborne electromagnetic surveys based on agglomerative hierarchical clustering. With this statistical classification method, the numerous 1D vertical resistivity profiles distributed over the area of interest can be summarized in a 2D horizontal map. Inverted electromagnetic soundings are aggregated into clusters according to their vertical resistivity profiles. These clusters define areas with a consistent vertical resistivity response, i.e. to the first order, areas with comparable vertical (hydro)-geological structure. The clustering method is illustrated with an extensive airborne electromagnetic survey conducted over Reunion Island. After a description of the method, we show that the proposed classification facilitates the confrontation between airborne geophysical results and geological or hydrogeological observations/data. We highlight that hierarchical clustering is of high interest for multi-scale studies, from regional to local scales. This approach introduces a new way to use geophysical surveys to map areas with specific geological/hydrogeological behaviors.