Evaluation of the errors and uncertainties resulting from urban canopy models and PBL schemes with physics-based ensemble

The urban forcing on energy and circulation can greatly influence the local evolution of the atmospheric boundary layer (BL) and air pollutant transport and dispersion. To accurately represent the urban effects, urban canopy models are widely used. However, the errors and uncertainties in predicting the depth and evolution of the urban BL and their impacts on air pollutant including greenhouse gases (GHG) concentrations are not well documented. This paper focuses on urban BL modeling, and aims to 1) quantify and compare the errors and uncertainties associated with different urban canopy models and PBL parameterizations; 2) investigate the impacts of the uncertainties brought by urban and PBL parameterizations on estimating GHG fluxes from Indianapolis. For these purposes, a moderately polluted period during 30 September to 2 October 2014 was studied using a 20-member physics ensemble based on WRF-ARW (Version 3.6.1). Each ensemble member is run with 3-nested domains of 13.5-, 4.5-, and 1.5-km horizontal grid space. The analysis based on the simulations of 1.5-km resolution shows that, in general, errors and uncertainties in temperature and wind predictions brought by PBL parameterizations are higher than that brought by urban canopy models. Also, the spatial uncertainties are larger than the temporal uncertainties for both wind and temperature for all the ensemble members. Inter-comparison between PBL parameterizations suggests that BOULAC PBL scheme may bring smallest errors and the MRF PBL scheme would induce the smallest spatial-temporal uncertainties for temperature prediction. Of all tested, the MYJ PBL scheme, however, has the best performance in wind. As for urban models, multi-layer urban canopy models (BEP and BEP+BEM) always introduce smaller errors and uncertainties than the single-layer urban canopy model (UCM). Based on the statistics, the MYJ PBL scheme combined with BEP+BEM urban canopy model are selected as the optimal parameterization for the predicting urban BL. The HYSPLIT dispersion model imported by the WRF physics-based ensemble meteorological fields is used as a proxy to study the uncertainties of pollutant concentrations related to the different urban and PBL parameterizations. The results indicate that the uncertainties in spatial distribution and maximum concentration caused by different PBL parameterizations would be much higher than those caused by different urban parameterizations.  These results can improve estimates of GHG fluxes in INFLUX.

Key words: physics-based ensemble, urban canopy model, air pollution