Particle fluxes over forests:: Analyses of flux methods and functional dependencies

[1] Particle number fluxes at two forests are computed using data from condensation particle counters and both the eddy covariance (EC) and relaxed eddy accumulation (REA) methods applied offline, post data collection. The results indicate that a 30-min integration period for flux calculation frequently applied for other scalars is also appropriate for particle number fluxes. Half-hour average particle number fluxes computed using REA and EC are statistically indistinguishable. Less than 1% of all fluxes derived using EC and REA differ by more than the mutual uncertainty bounds, and the average ratio of fluxes computed using the two approaches is 1.00 +/- 0.01. This finding tends to support assertions that, for the size of particles under study (with diameters < 100 nm), the assumption of similarity with heat transfer is likely robust. Data from both sites indicate higher particle deposition velocities (v(d)) for particle ensembles with lower number geometric mean diameter (GMD), for GMD of 10 to 100 nm, although the decrease with increasing GMD is less marked than is predicted by the deposition models such as that of Slinn. The relationships between median and mean vd conditionally sampled by GMD are well described by power law fits. Particle vd exhibits a linear dependence on friction velocity for the range of conditions studied and also enhancement under unstable conditions. Similarities between average particle vd conditionally sampled by number GMD, friction velocity, and stability from the two forest sites point to the generalizability of the empirical fits derived herein.