Abstract

To investigate small-scale transport and dispersion (T&D) within urban areas, we couple a large-eddy-simulation (LES)-based urban-scale fluid solver (EULAG), with the mesoscale Weather Research and Forecasting (WRF) system. The WRF model uses two different urban canopy models (UCM) to parameteriz... Show moreTo investigate small-scale transport and dispersion (T&D) within urban areas, we couple a large-eddy-simulation (LES)-based urban-scale fluid solver (EULAG), with the mesoscale Weather Research and Forecasting (WRF) system. The WRF model uses two different urban canopy models (UCM) to parameterize urban effects: a single-layer parameterization (SLUCM) and a multilayer building-effect parameterization (BEP) model coupled to Bougeault and Lacarrère planetary boundary-layer scheme. In contrast, EULAG uses the immersed-boundary (IMB) approach to explicitly resolve complex building structures. Here we present details of the downscaling transfer approach where the mesoscale conditions are used to supply initial and lateral boundary conditions for EULAG. We demonstrate its benefits and applicability to solve dispersion problems in the complex urban environment of Oklahoma City. The coupled modeling system is evaluated with data obtained from two intensive observation periods (IOP) of the Joint Urban 2003 experiment, representative for daytime convective (IOP6) and nighttime stable (IOP8) conditions. We assess the sensitivity of urban dispersion simulations to accuracy of the WRF-generated mesoscale conditions. The results show that WRF-BEP reproduces the observed mean near-surface and boundary-layer wind and temperature fields during daytime conditions, and provides accurate statistics during the nighttime more accurately than WRF-SLUCM. The EULAG model performance is exhibited with time-averaged and instantaneous peak concentration statistics. The improved statistics during IOP6 are achieved by using WRF-BEP indicating how important the proper meteorological conditions are to the accuracy of small-scale urban T&D modeling. Show less