Multivariate Strategy for Understanding Soil Features from Rare-Earth Element Profiles: A Focus on Data Normalization

The interest in assessing the behavior of rare earth elements (REE) in natural environment is constantly increasing due to their numerous applications in both environmental and technological fields. However, current methodologies for analyzing REE distributions are based on normalization of REE concentration profiles to lithological values, potentially resulting in different outcomes depending on which lithological values are used for normalization, affecting the interpretability of the data. The present work proposes an alternative approach for analyzing REE concentration profiles by applying principal component analysis (PCA) to create REE chemometric maps. Data compression allows the visualization of REE behavior using a red-green-blue (RGB) color scale (PC1 = red; PC2 = green; PC3 = blue) directly on a geographical map, maintaining the information associated with the chemical properties of rare earth elements. This highlights similarities and differences in the compositional REE distribution of natural soils, facilitating the interpretability of REE data and potentially leading to new insights related to seemingly unrelated samples. Additionally, PCA applied to soil variables correlates with REE patterns, and provides deeper insights into soil properties in an unsupervised manner, enhancing the interpretation of soil characteristics and implementing the ability to monitor environmental changes and study soil evolution processes. Of particular significance is the fact that applying the proposed methodology to non-normalized data yields results consistent with those derived from normalized datasets. This demonstrates that this approach does not only overcome normalization challenges but also supports the classical approach from a new methodological perspective, paving the way for broader applications.