Self-organising feature map (SOFM) algorithms applied to manganese mineralisation in soils close to an abandoned manganese oxide mine

This paper proposes a multi-level self-organising map (SOFM) approach in studying manganese minerals interdependence in soils close to an abandoned Mn oxides mine. Multiple SOFM algorithms for data clustering were applied on Mn minerals identified by X-Ray diffractometry contained in four hundred soil samples from the periphery of the abandoned mine. Emerging structures from the Mn minerals (bixbyite, cryptomelane, ramsdellite, pyrolusite and braunite) were analysed using SOFM and two of the minerals (cryptomelane and braunite) were found to be influential in cluster formation. The findings of the study demonstrate the suitability of data mining in characterising Mn minerals interdependence in soils close to the abandoned Mn oxides mine and highlight the underlying, issues of which applicants of the method need to be aware of.