Spatial Factors Influencing Winter Primary Particle Sampling and Interpretation

ABSTRACT

Aerosol samplers collect material that is locally generated as well as that transported from upwind; knowing the extent of the area from which the sample is drawn is necessary for proper interpretation of sampler data. The U.S. Environmental Protection Agency (EPA) PM_2.5 monitoring guidelines recognize a conceptual hierarchy of sampler spatial representation, but provide no objective measures of a site’s spatial representativeness. A case study of a sampler tributary area in central California provides insights into the factors that determine a sampler’s spatial representation. Winter diurnal cycles of fine particle concentrations at places of habitation ranging from urban cores to small farm towns show a marked cycle that can be linked to local human activity. Assessment of the possible causes of the observed cycles leads to the hypothesis that local sources dominate primary particle mass in winter samples. The hypothesis was tested using a simple model to relate routine 24-hr PM₁₀ and PM_2.5 samples to a sampler’s surroundings. Model results indicate that even minor sources very close to a sampler will overwhelm any regional component in a sample. The results for the cases studied also demonstrate that, in winter, most coarse (PM_10-2.5) particles collected are less than 2 hr old, and most primary fine (PM_2.5) particles are less than 4 hr old. Even on days that are not truly “stagnant,” samplers are very strongly influenced by their immediate surroundings (distances less than 10 km), and only weakly influenced by regional emissions.

The implications for interpretation of sample analyses are as follows: 1. Typical PM sampling networks are unlikely to represent regional conditions;

2. Similarity of samples in time and space between widely separated samplers probably arises from sampling analogous local environments rather than a uniformly mixed regional air mass;

3. Even weak sources near a sampler will prevent regionally representative samples, so that “background” specification in models can be strongly skewed by misapplication of sampler data;

4. Source-receptor relationships within a single modeling grid cell can cause measured and modeled source impacts at a sampler to diverge by orders of magnitude, even for grid cells as small as 1 km; and

5. Differential deposition of coarse and fine particles will skew source apportionment by chemical tracers unless the tracers and the source emissions have the same size distribution.