Illicit Sewage Connection Fuels CH₄ and CO₂ Emissions in Storm Sewer: Tracking from Sources to Mechanical Influences

The complexity of spatiotemporal CH4 and CO2 patterns caused by illicit connections introduces substantial uncertainty into accounting frameworks for greenhouse gases (GHGs) in urban sewer systems. Using dissolved organic matter (DOM) fluorescence fingerprints (FFs) as tracers, a novel diagnostic tool integrating the Bayesian mixing model (MixSIAR) and isolation forest machine learning was developed in this study to (1) quantify the probabilistic contributions of organic carbon (OC) from domestic sewage (DS) and storm runoff (SR) sources and (2) geolocate DS connection sites in storm sewers. We employed multivariate analysis to reveal the influence of illicit connections on sewer CH4 and CO2 emissions. Trp-b and Co-r were the optimal FF tracers for tracking DS and SR. Sites 5 and 8 were identified as the high-probability connection hotspots exhibiting predominant DS-derived OC contributions (51.51% and 52.6%). We observed a power-law response of the CH4 and CO2 emissions to illicit connections, primarily influenced by dissolved organic carbon (DOC), DOM compositions, flow rate, temperature, dissolved oxygen (DO), and factors related to sediment nitrogen. This study is critical in addressing the challenges of establishing practical tools to diagnose illicit connections and define their influence mechanisms and the field GHG boundary for robust accounting in the sewer.