Assessing geochemical anomalies using geographically weighted lasso

Geochemical patterns in the surficial media are complex because of the various processes involved in their formation. These processes may or may not relate to mineralization. Therefore, geochemical anomalies identified are usually subject to uncertainty. Geographically weighted lasso (GWL) was adopted in this study to model the non-stationary relationships between the geological/geographical features and geochemical patterns in the surficial media. The sparsity and reliability of regression coefficients in the GWL modeling renders it suitable for quantifying the controls that geological/geographical features have on geochemical patterns. This information can improve current knowledge about the complexity of geochemical patterns. A case study in which the GWL model was established between geological/geographical variables and the geochemical patterns of Cu observed in stream sediments in the Southwestern Fujian province of China was presented. The following findings were obtained: (1) geological features show a major control on the geochemical patterns of Cu in the studied area. However, the geochemical patterns of Cu have been modified to some extent by secondary processes, resulting in the relationships between geochemical signatures and mineralization becoming more complex and uncertain, and (2) the GWL modeling can provide useful information for evaluating geochemical anomalies, which can help reduce the uncertainty and benefit geochemical exploration at a later stage.