Optimization of geochemical anomaly detection using a novel genetic K-means clustering (GKMC) algorithm

Due to the complicated characteristics of regional geochemical data from stream sediments as a result of the complexity of geological features, detection of multi-elemental geochemical footprints of mineral deposits of interest is a challenging task. As a way to address this, a hybrid genetic algorithm-based technique, namely genetic K-means clustering (GKMC) algorithm, is proposed here for optimum delineation of multi-elemental patterns (both anomaly and background) in stream sediment geochemical data. To do so, factor analysis and sample catchment basin modeling were coupled with GKMC and traditional K-means clustering (TKMC) methods for identification of anomalous multi-elemental geochemical footprints of deposits of porphyry copper and skarn copper in the 1:100,000 scale Varzaghan map sheet, northwest Iran. Based on higher prediction rate, it can be inferred that the model derived by GKMC is superior to the one derived by TKMC. In addition, the strong anomaly classes of the GKMC and TKMC models predict, respectively, similar to 83% and similar to 66% of the porphyry/skam Cu deposits in similar to 22% and similar to 36% of the study district. Thus, the geochemical anomaly targets derived by the GKMC method are more reliable than those generated by the TKMC method. This revealed that the GKMC algorithm is an efficient and robust tool for recognizing multi-element geochemical anomalies for mineral exploration.