Dustfall source-apportionment and source-oriented health risk assessment using unsupervised machine learning, PMF,...

In arid regions such as Xinjiang, frequent dust storms, sparse vegetation, and low precipitation result in persistently high atmospheric dust loads. Atmospheric dustfall acts as a critical carrier for heavy metal (HM) pollutants, this environment not only facilitates the long-range transport and enrichment of heavy metals (HMs) in dustfall but also leads to prolonged human exposure through inhalation and ingestion, making accurate source identification and health risk assessment particularly critical for public health protection. In this study of 40 dustfall samples from the Hotan Oasis, a hotspot area in Xinjiang of China, were collected and an integrated framework, combining the self-organizing map (SOM) and K-means clustering algorithms (SOM-K), positive matrix factorization (PMF), Hybrid Single Particle Lagrangian Integrated Trajectory (HYSPLIT) model, and source-oriented risk assessment, was developed to quantify the sources and probabilistic source-oriented health risk assessment of HMs in atmospheric dustfall. The results indicate that HMs in the dustfall originated from five main sources: industrial activities, road traffic emissions, mining and coal burning, wind-sand dust, and a mixed source of livestock manure and fertilizer application, with anthropogenic activities contributing more than natural sources. Dustfall was also influenced by long-distance transport, predominantly from the Taklimakan Desert, accounting for 94.57 % of the trajectories. Health risk assessment further revealed a distinct risk pattern: children faced higher non-carcinogenic risks (THI _adults= 1.66E-01, THI _children= 3.34E-01), whereas adults were more susceptible to carcinogenic risks (TCR _adults = 2.44E-04, TCR _children = 8.08E-05). The source-oriented assessment linked sources to risks, identifying mining and coal burning as the largest contributors to carcinogenic risk, traffic emissions as the main source of non-carcinogenic risk, and As, Pb, Zn, Cd as the primary risk drivers. This study demonstrates the power of an integrated approach in unraveling complex pollution pathways and provides a scientific basis for formulating targeted strategies to mitigate health risks in arid oasis environments and similar regions globally.