Assessing Indices Tracking Changes in River Geochemistry and Implications for Monitoring

In geochemical data analysis, assessing the potential of new techniques to identify compositional time-space changes is of great interest for monitoring purposes. This work aims to evaluate, in the light of the compositional data analysis perspective, the performance of different statistical indices in tracing the evolution of a geochemical composition and the relationships among its parts. To reach this goal, source-to-sink chemical changes in water and stream sediment composition of the Tiber river (central Italy) are analyzed using three indices: (i) the cumulative sum of unclosed perturbation factors of each composition (row sum) with respect to a reference composition; (ii) the robust Mahalanobis distance, describing the compositional differences from the same reference and, (iii) the geometric mean of each composition as a measure able to capture the interactions among the parts. The results highlight the major compositional changes downriver, allowing to explore geochemical footprints' propagation and their natural or anthropogenic origin. The tested indices are consistent in most cases, particularly if high-variability species are treated separately and low values are rare. Under this latter condition, the geometric mean of the composition shows a close connection with the cumulative sum of unclosed perturbation factors. This indicates that both indices inherit the complex history of the changes, well capturing the interactions among the parts under the influence of environmental drivers. With this awareness, the application of these methods in monitoring and applied geochemical studies could offer new insights into the inner workings of river systems and their resilience to environmental pressures.

Automated summary

Assessing the potential of new techniques to identify compositional time-space changes is important in geochemical data analysis for monitoring purposes. This study evaluates the performance of different statistical indices in tracing the evolution of geochemical composition and the relationships among its parts. The results show that the tested indices are consistent in most cases and can capture the interactions among the parts under the influence of environmental drivers.