Source apportionment of ubiquitous organophosphate esters (OPEs) in soil is complicated by high spatial heterogeneity, which often causes conventional Positive Matrix Factorization (PMF) models to produce unstable and physically implausible solutions, particularly in pollution hotspots such as the Yangtze River Delta (YRD). To overcome this limitation, we developed a novel spatially constrained PMF (SC-PMF) model that incorporates geographical information as a penalty term within the PMF objective function. Applied to a comprehensive soil data set from the YRD, the SC-PMF model successfully resolved six robust and physically plausible sources, in stark contrast to the conventional PMF, which failed to produce a stable solution under Bootstrap analysis. The identified sources included not only expected inputs such as agricultural activities but also a suite of previously obscured pathways, revealing contributions from consumer plastic packaging, long-range atmospheric deposition, and informal waste disposal. Critically, the model also decoupled distinct industrial processes by differentiating primary manufacturing emissions from secondary pollution caused by the degradation of industrial additives. These results prove that incorporating spatial constraints is a powerful strategy for resolving source ambiguity in heterogeneous media. The SC-PMF model thus offers significant methodological advancement for accurately tracing contaminant pathways and informing targeted pollution management.
Zhao et al. (Fri,) studied this question.