Revisiting dry deposition modelling of particulate matter on vegetation at the microscale

Abstract

Dry deposition is an important process determining pollutant concentrations, especially when studying the influence of urban green infrastructure on particulate matter (PM) levels in cities. Computational fluid dynamics (CFD) models of PM capture by vegetation are useful tools to increase their applicability. The meso-scale models of Zhang et al. (Atmos Environ 35:549–560, 2001) and Petroff and Zhang (Geosci Model Dev 3(2):753–769, 2010) have often been adopted in CFD models, however a comparison of these models with measurements including all PM particle sizes detrimental to health has been rarely reported and certainly not for green wall species. This study presents dry deposition experiments on real grown Hedera helix in a wind tunnel setup with wind speeds from 1 to 4 m s$^{-1}$ and PM consisting of a mixture of soot (0.02 - 0.2 $\upmu$m) and dust particles (0.3 - 10 $\upmu$m). Significant factors determining the collection efficiency (%) were particle diameter and wind speed, but relative air humidity and the type of PM (soot or dust) did not have a significant influence. Zhang’s model outperformed Petroff’s model for particles < 0.3 $\upmu$m, however the inclusion of turbulent impaction in Petroff’s model resulted in better agreement with the measurements for particles > 2 - 3 $\upmu$m. The optimised model had an overall root-mean-square-error of $\sim$ 4% for collection efficiency (CE) and 0.4 cm s$^{-1}$ for deposition velocity ($v_d$), which was shown to be highly competitive against previously described models. It can thus be used to model PM deposition on other plant species, provided the correct parameterisation of the drag by this species. A detailed description of the spatial distribution of the vegetation could solve the underestimation for particle sizes of 0.3 - 2 $\upmu$m.

Appendices

Appendix A: Conditions of the incoming air stream

The table below provides the meteorological conditions of the incoming air stream measured at the inlet of the plant section, for an empty wind tunnel system (i.e. blanco) and for a wind tunnel with Hedera helix.

Table 7 The wind speed, U$_{in}$ (m s$^{-1}$), relative humidity, RH ($\%$), and temperature, T ($^\circ$C), of the incoming air stream, measured at the inlet of the plant section, for an empty wind tunnel system (i.e. blanco)

Table 8 The wind speed, U$_{in}$ (m s$^{-1}$), relative humidity, RH ($\%$), and temperature, T ($^\circ$C), of the incoming air stream, measured at the inlet of the plant section, for a wind tunnel system with H. helix

Appendix B: Grid convergence study

The table below shows the results of the grid convergence study of the present study.

Table 9 Outcome of a grid convergence study of the incoming mean wind speed ($U_{in}$) for Beaufort classes 1 to 3, in terms of DE (%) and GCI (%) relative to the finest grid

Appendix C: Deposition velocity of dust particles on leaves of H. helix

The following figures show the results, in terms of the deposition velocity (v$_d$, m s$^{-1}$) of different models used to simulate PM deposition onto leaves of H. helix.

Fig. 11

Measured total $v_d$ (m s$^{-1}$, bullets) and modelled deposition velocity (m s$^{-1}$, bar plot) of each PM deposition mechanisms (1$^{st}$ bar = Petroff model, 2$^{nd}$ bar = Zhang model) for an incoming PM$_{10}$ concentration of 100 (left) and 150 $\upmu$g m$^{-3}$ (right)

Fig. 12

Measured total $v_d$ (m s$^{-1}$, bullets) and modelled deposition velocity (m s$^{-1}$, bar plot) of each PM deposition mechanisms with the modified model combining the model of Zhang and Petroff for an incoming PM$_{10}$ concentration of 50 (A), 100 (B) and 150 $\upmu$g m$^{-3}$ (C)

Fig. 13

Measured total $v_d$ (m s$^{-1}$, bullets) and modelled deposition velocity (m s$^{-1}$, bar plot) of each PM deposition mechanisms with the modified model with a turbulent Stokes number for an incoming PM$_{10}$ concentration of 50 (A), 100 (B) and 150 $\upmu$g m$^{-3}$ (C)