Simulated Particle Vertical Transport by Mesoscale Convective Systems and its Dependency on Grid Spacing

Mesoscale convective systems (MCSs) can produce enormous rainfall and also transport air downward via convective and mesoscale downdrafts. Previous observational studies in the Amazon show that the concentration of ultrafine particles is relatively high in the upper troposphere and that free tropospheric ultrafine particles may also be transported into the boundary layer via convective downdrafts. However, it is unclear how efficiently aerosol particles are transported vertically in deep convective systems between different altitudes and how important such systems are for vertical transport in general. Because the spatiotemporal coverage of vertical velocity and aerosol measurements are very limited, we use modeling to investigate vertical transport since both kilometer-scale and finer resolution models have shown promise in reproducing essential MCS features. Nonetheless, model grid spacing can significantly impact the model representation of in-cloud vertical motion. Due to expensive computational cost, atmospheric chemistry models are often configured using a relatively coarse grid (>10 km) using convective parameterization that could produce unrealistic convective transport in simulated MCSs.

To understand the implication of model grid spacing on vertical transport of particles within MCSs, we use the Weather Research and Forecasting model coupled with chemistry (WRF-Chem) to enable tracer simulations in a range of horizontal grid spacings (10 km, 4 km, 1 km, and 250 m). The results of simulated MCS events are compared against observations collected during the Green Ocean Amazon (GoAmazon) field campaign, including precipitation radar-retrieved reflectivity and vertical velocity. The MCS lifecycle is more realistically simulated by the 4 km simulation than 10 km with a primary difference being explicit representation of convection in the 4-km setup and parameterized convection in the 10-km setup. Moreover, the simulated in-cloud vertical motion distributions vary significantly among the four resolutions. While the maximum updraft and downdraft speeds in the 1 km simulation are double those in the 4 km run, the difference between 1 km and 250 m speeds are relatively small. The vertical velocity distribution simulated by the 250 m simulation compares best with observations. Although the downward transport from the upper troposphere (UT) to the boundary layer (BL) is greatly enhanced when using a higher model resolution, it is much less efficient than upward transport from the BL to the UT. We found it is more likely aerosols in the UT may be transported down to middle troposphere (MT) through mesoscale descent, and the negative buoyant convective downdraft driven by heavy precipitation evaporation cooling facilitate transport from the MT to the BL. For both upward and downward transports, results show the conventional atmosphere-chemistry coupled model likely underestimates the rate of aerosol vertical transport. This suggests the effect of MCS induced vertical motions should be represented when exploring mechanisms such as BL secondary aerosol formation in relation to ultrafine aerosol concentration in the Amazonian rainforest.