The presence of naturally-occurring dust is a conspicuous meteorological phenomenon characterised by very high load of relatively coarse airborne particulate matter (PM), which may contain also various deleterious chemical and biological materials. Much work has been carried out to study the phenomenon by modelling the generation and transport of dust plumes, and assessment of their temporal characteristics on a large (>1000 km) spatial scale. This work studies in high spatial and temporal resolution the characteristics of dust presence on the mesoscale (>100 km). The small scale variability is important both for better understanding the physical characteristics of the dust phenomenon and for PM exposure specification in epidemiological studies. Unsupervised clustering-based method, using PM10 records and their derived attributes, was developed and applied to detect the impact of dust at the observation locations of a PM10 monitoring array. It was found that dust may cover the whole study area but very often the coverage is partial. On average, the larger the spatial extent of a dust event, the higher and more homogeneous are the associated PM10 concentrations. Dust event lengths however, are only weakly associated with the PM concentrations (Pearson correlation below 0.44). The large PM concentration variability during spatially small events and the fact that their occurrence is strongly correlated with the elevation above sea level of the reporting stations (Pearson correlation 0.87, p-value < 10-5) points to small scale spatiotemporal dynamics of dust plumes.
- Cluster analysis
- Dust presence detection
- Mesoscale character of dust plumes