Development of a Multi-Scale Modeling Framework for Urban Simulations in the Weather Research and Forecasting Model

Downscaling meteorological simulations from the meso-scale (Δx > 2km) to the micro-scale (Δx < 100m) is a difficult challenge with no clear “best” solution. Commonly used meteorological models, such as the Weather Research and Forecasting (WRF) model, have been designed for meso-scale simulation and lack features necessary for micro-scale simulation, such as flexible gridding procedures. Here, we develop a multi-scale framework within the WRF model that enables simulations of flow over urban terrain using realistic time-varying lateral boundary conditions nested down from meso-scale simulations.

The multi-scale framework depends on the use of two major code developments. First, the ability to refine the vertical grid between concurrently run nested domains is used to created nested simulations that span grid resolutions from 6.05 km to 2 m while maintaining appropriate grid aspect ratios. Second, the immersed boundary method (IBM) of Lundquist et al. 2012 is improved and used for resolving flow around urban terrain. This improved IBM algorithm can be applied to a domain nested within a traditional meso-scale WRF domain that uses a terrain-following coordinate. Additionally, the new IBM algorithm includes a log-law boundary condition. This framework enables the WRF-IBM model to be used for investigations into downscaling and multi-scale modeling over complex topography, such as urban terrain.

Validation of the WRF-IBM model is performed using data from the first continuous tracer release of IOP-3 in the Joint Urban 2003 Field Campaign in Oklahoma City. Simulations of the tracer release are completed using both idealized and real lateral boundary conditions. The idealized configuration is a two-domain nested setup where turbulence is developed on a flat parent domain with periodic lateral boundary conditions before the initialization of a nested domain with the urban topography. The real configuration includes five nested domains with grid resolutions ranging from 6.05 km to 2 m and the outermost domain forced using NARR data. Performance metrics evaluating model skill show improvements using the new IBM algorithm versus previously published WRF-IBM simulations.