Dynamic tracking of multidimensionally annotated chemical fingerprints in non-targeted screening: a molecularly resolved method for contaminant source tracking.

Environmental transformations and multi-source mixing of organic contaminants (OCs) in real river systems substantially complicate reliable source identification during downstream transport. This study proposes a molecularly resolved source-tracking method within a non-targeted screening (NTS) framework that integrates multidimensional chemical fingerprinting with along-stream dynamic tracking, enabling stable source identification under complex transport and transformation conditions. By organizing compounds into chemically coherent molecular families, the proposed multidimensional fingerprinting approach amplifies chemical information, expanding 162-929 source-specific compounds into 253-1733 structurally and transformation-resolved attribute entries per source. This expansion allows source information to propagate downstream despite continuous molecular-level structural evolution. Across the river network, the method continuously tracked source-associated chemical signals and generated spatial contribution patterns that converged with those obtained through least-squares similarity analysis, while retaining compound-level and process-level interpretability. Compared with conventional static, compound-based tracking approaches, the number of traceable compounds increased by approximately 4.61-fold (from 23 to 106). Notably, effective source discrimination was maintained even when the chemical compositions of OCs from different sources were highly similar. A chemically interpretable and process-oriented source-tracking strategy is presented, advancing NTS from static detection toward dynamic analysis and providing a feasible and robust pathway for contaminant source identification in complex riverine systems.