An extended local gap statistic for identifying geochemical anomalies

• A hybrid of gap statistic and sliding window technique is presented.
• Probability is considered to further refine significant anomalies.
• A case study demonstrates the advantages of the local gap statistic method.

In this study, a hybrid of gap statistic and sliding window technique, termed a local gap statistic, was presented to estimate the threshold and associated significance. The hybrid method can not only detect local anomalies in the varied geochemical background, but also consider the significance of the derived threshold, in that the anomalous classification of some locations is rejected if the computed probability is larger than the predefined one. A case study which processes a geochemical dataset from soil samples obtained in Inner Mongolia of China was carried out to test the method. The results indicate that (1) the local gap statistic is a useful tool to identify geochemical anomalies from varied background; and (2) the derived probability not only serves to further refine anomalies, but also provides an assessment of the uncertainty involved in delineated anomalies.