In air quality forecasting, meteorological variables are of great importance. In particular, the planetary boundary layer(PBL) is one of the most important factors influencing the vertical distributions of air pollutants. As PBL is characterized by th...
In air quality forecasting, meteorological variables are of great importance. In particular, the planetary boundary layer(PBL) is one of the most important factors influencing the vertical distributions of air pollutants. As PBL is characterized by the complex turbulence precesses in the meteorological model, it should be parameterized by determining the turbulent properties. Therefore model results of pollutant concentrations are highly dependent on parameterization methods. In this study, as a case study, we carried out three sensitivity tests from different PBL turbulence parameterization schemes: nonlocal(YSU), local(MYJ), hybrid(ACM2) for WRF-CMAQ, and detailed statistical analysis for the pollutants were carried out over day and night separately; daytime is characterized by thermal turbulence whereas nighttime is dominant by mechanical turbulence.
The results showed that temperature and relative humidity were simulated relatively better for both day and night except for some meteorological variables such as wind speed. Wind speed was overall overestimated by all parameterization schemes in the model; therefore, it is necessary to improve wind speed simulation capability in simulating the air pollutants by WRF model. Relatively, non-local PBL scheme outperforms local scheme during daytime; local scheme simulates more than twice as much as observations during daytime. It is therefore indicating that the non-local method that parameterizes the turbulence with the consideration of overall vertical profile is relatively more appropriate for simulating wind speed. PBL height(PBLH) was found to be superior to the non-local scheme during daytime when large eddies were dominant and buoyancy is well associated, while local scheme shows better capabilities during nighttime when turbulence such as small eddies and molecular diffusion were mainly caused by shear. PM2.5 concentrations were underestimated at all times, while non-local scheme simulated relatively higher concentrations among the three schemes, thereby yielding simulations close to the observation. Local scheme showed relatively better capability in PBLH simulations in nighttime, whereas it showed the worst capability in PM2.5 simulation presumably due to overestimated wind speed.
As a result of modeling analysis for day and night separately, it is advantageous that a hybrid technique would be recommended to parameterize turbulence characteristics by applying the non-local method during the day and the local method at night to improve the simulation capability of atmospheric boundary layer parameterization over the Seoul metropolitan area. Further, it would be also necessary to apply model sensitivity studies to various cases and other urban areas to provide a legitimate, integrative, interpretive evaluation in association with PBL schemes in South Korea.