Simulations to support local source apportionment using forward and inverse simulations of urban dispersion at the micro-scale

Abstract:

Air pollution simulations are typically performed at meso-scale resolutions and micro-scale urban features are parameterized or neglected. Micro-scale impacts on airflow and pollutant transport can be significant at the urban scale where neighborhood air quality is sensitive to the strength and distribution of local air pollution sources. In particular, identifying the source of local air pollution is unrealistic at typical 3 km resolutions when the urban area varies at much finer scales. In the current study, we use the Weather Research and Forecast (WRF) model and the HYbrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model to investigate neighborhood-scale pollutant transport using nested meso-scale to micro-scale domains. The region of interest is West Oakland, CA, the site of previous observational campaigns. The WRF model is used to simulate airflow with resolutions ranging from ~3 km down to the building scale (2-50 m). Simulations with grid spacing less than 10 m are performed using an immersed boundary method (IBM) to represent buildings in WRF. This WRF-IBM model is able to resolve building surfaces and capture flow structures around them. Both USGS 1/3 arc-second DEM data (~10 m) and NLCD 1 arc-second land-cover data (~30 m) are implemented into WRF for the current simulations, in addition to detailed building footprints. Meteorological observation data for the selected time period are used to validate the forward WRF simulations. The velocity fields from WRF are then taken by HYSPLIT to perform both forward and backward Lagrangian modeling, exploring the transport of passive scalars over the urban area at different grid resolutions. The results of the forward modeling show clear differences not only in terms of the micro-scale flow structure but also in the mean transport direction and plume width from a synthetic source location. The footprint obtained from the inverse modeling shows that the source area is sensitive to both the spatial scales being resolved and the selected time periods. This meso-scale to micro-scale modeling framework allows detailed representation of urban areas in air pollution models, thus enabling local source apportionment techniques and informing the use of trajectory and dispersion analysis for applications at different scales.