Differential dissolution-induced concentration drift in sediment fingerprinting: Theory, validation, implications, and resolution

Abstract

Reliable identification of sediment sources is fundamental to understanding erosion dynamics and implementing targeted watershed management. Sediment fingerprinting has therefore been widely applied to quantify sediment provenance. Here, we identified a previously overlooked process, termed Differential Dissolution-Induced Concentration Drift (DDICD), whereby element-specific dissolution during transport induces systematic shifts in sediment geochemical composition. This process implies that post-transport elemental concentrations depend not on absolute elemental properties, but on their relative differences with respect to coexisting elements, potentially challenging the concentration-conservation assumption underlying conventional sediment fingerprinting techniques when DDICD occurs. To address this issue, we developed a theoretical framework for DDICD and formulated a ratio-tracer-based correction approach. Theoretical analyses using synthetic datasets revealed that DDICD intensity was primarily controlled by the heterogeneity of elemental dissolution rates and initial masses, as well as by the rank concordance between these properties, originating within individual sediment sources and accumulating through source mixing to amplify sediment fingerprinting errors. Empirical results from the Yangtze River Basin showed that concentrations of highly soluble elements generally decreased during transport (e.g., Ca and Mg decreased by 15.02% and 8.06%, respectively), whereas relatively insoluble elements generally increased (e.g., Al increased by 6.39%). Sediment unmixing results further indicated that the mean absolute error (MAE) of source contribution estimates increased logarithmically with transport distance and elevation drop. Compared with element tracers, the use of element ratios substantially improved model accuracy and robustness, reducing mean MAE and its standard deviation by 9.83 and 0.45 percentage points, respectively. All these findings provided strong evidence for the occurrence of DDICD in natural river systems and demonstrated its substantial influence on sediment fingerprinting results. Consequently, DDICD should be explicitly considered in sediment fingerprinting of large catchments or regions characterized by intense weathering and abundant carbonate or silicate minerals, as neglecting this process may introduce systematic biases into sediment source apportionment.