Abnormal industrial wastewater discharge into urban sewer networks can disrupt the operation of wastewater treatment plants (WWTPs) and increase pollution risks to receiving water bodies. However, large-scale deployment of total phosphorus (TP) chemical monitoring at numerous sewer nodes is costly and difficult to maintain for high-frequency data acquisition, limiting rapid anomaly identification and source tracing. This study develops a hierarchical online monitoring and spatiotemporal tracing framework using the service area of the Xiani WWTP in Dongguan, China, as a case study. The study area was divided into six monitoring zones and 169 drainage blocks, with 57 online monitoring stations deployed at trunk and branch/secondary pipelines. Class-A stations directly monitored TP, while Class-B stations continuously monitored electrical conductivity (EC), pH, and liquid level. A “plant–network–zone–block–source” association framework was established by integrating pipe network topology, flow direction, and estimated travel time. Random forest-based contribution analysis, Fréchet distance trajectory matching, and Bayesian-Markov Chain Monte Carlo (MCMC) inference were combined to progressively rank candidate nodes, drainage blocks, and discharge sources. Results showed a moderate positive correlation between TP and EC (Spearman’s $\rho$ = 0.50, $p$ $<$ 0.05, $n$ = 116), indicating that EC can serve as an auxiliary indicator for screening phosphorus-containing industrial wastewater inputs but cannot replace direct TP measurement. In two verified abnormal discharge events, EC changes at Station 19B and Station 17B preceded TP anomalies at the WWTP influent by approximately 4 h and 6 h, respectively, consistent with estimated flow travel times. Field investigations further identified abnormal industrial wastewater discharge clues in the corresponding areas. Following targeted inspections, enterprise rectification, and enforcement actions, TP concentration at the A-Dapu Station decreased by 50%. The study demonstrates that the proposed hierarchical monitoring and multi-model tracing framework can progressively narrow anomaly investigation from the WWTP service area to candidate pipelines, drainage blocks, and discharge sources, supporting abnormal discharge screening and digital drainage management in industrially dense urban areas.