Sequential Factor Analysis as a new approach to multivariate analysis of heterogeneous geochemical datasets: An application to a bulk chemical characterization of fluvial deposits (Rhine–Meuse delta, The Netherlands)

Abstract

Sequential Factor Analysis (seqFA) is presented here as an enhanced alternative to multivariate factorial techniques including robust and classical Factor Analysis (FA) or Principal Component Analysis (PCA). A geochemical data set of 145 sediment samples from very heterogeneous, mainly riverine, deposits of the Rhine-Meuse delta (The Netherlands) analyzed for 27 bulk parameters was used as a test case. The innovative approach explicitly addresses the priority issues when performing PCA or FA: heterogeneity and overall integrity of the data, the number of factors to be extracted, and which optimum minimal set of key variables to be included in the model. The stepwise decision process is based on quantitative and objectively derived statistical criteria, yet also permitting arguments based on geochemical expertize. The results show that seqFA, preferably in combination with robust methods, yields a highly consistent factor model, and is favorable over classical methods when dealing with heterogeneous data sets. It optimizes rotation of the factors, and allows the extraction of less distinct factors supported by only a few variables, thus uncovering additional geochemical processes and properties that would easily be missed with other approaches. The identification of key variables simplifies the geochemical interpretation of the factors, and greatly facilitates the construction of a geochemical conceptual model. For the case of the fluvial deposits, the conceptual model effectively describes their bulk chemical variation in terms of a limited number of governing processes.