Application of Euler Deconvolution to Magnetic Data for Mineral Exploration in the Arabian Shield

Euler Deconvolution is a widely utilized depth determination method that showed effectiveness characterizing geological model consisted on structures such as dykes, sills, faults and other simplified geometries showing lateral changes in magnetic susceptibility. The method's least-squares windowed approach renders its solutions more robust in comparison to similar methodologies. However, the price to pay for this robustness is the well-known scattering from the inversion process, commonly referred to in literature as "sprayed solutions". In this study, we apply this to synthetic data to examine the scattered behavior, as well as a traditional filter that relies on the expertise of the interpreter. To address this challenge, we propose an unsupervised machine learning algorithm for the generated solutions, facilitating a precise estimation of the locations of underlying magnetic sources. A case study in the Arabian Shield was utilized for this approach to mineralized deposits. A profile, averaging 2 km in length, was strategically aligned perpendicular to the strike direction of the surface manifestations of the mineral deposits to optimize the detection of subsurface structures. The analysis of magnetic anomalies reveals a complex geological setup on the western side of the mineralized outcrop. Most of these anomalies are linked to sub-vertical sulfide bodies located near the surface. A qualitative assessment of the positive magnetic anomalies indicates the presence of magnetite-rich intrusions, possibly connected to porphyry systems. Conversely, negative anomalies detected over individual sulfide deposits and a broader magnetic low zone suggest influences beyond simply mound-related anomalies. The raw standard Euler method yields scattered results, making it challenging to determine the extent of geological body clustering from the magnetic anomaly data. However, applying a k-means clustering algorithm to these scattered outputs produced a reasonable estimate of the number of geological bodies responsible for those anomalies. Overall, the results from the Euler method suggest that mineralized deposits are associated with linear features located at depths ranging from the surface to 10 meters. This study illustrates that combining the Euler deconvolution method with machine learning-driven techniques benefits mineral exploration due to its cost-effectiveness and rapid execution. This represents a significant advancement in mining exploration, particularly in identifying and characterizing subsurface mineral deposits during the greenfield stage, especially in the Arabian Shield and similar geological environments.