Developing an integrated urban modeling system for the WRF model

Fei Chen

NCAR/RAP
Abstract:

Rapid expansion of urban areas caused many adverse effects on air quality, energy and water supply/demand, and emergency response. Meanwhile, today’s numerical weather prediction (NWP) models run with grid spacing of 1 km, and at such fine scales, the role of urban land-use in local and regional weather needs to be represented. It is important for NWP models to capture the effects of urban areas on wind, temperature, and humidity in the boundary layer and their influences on the boundary layer depth. Air dispersion and quality models will benefit from improved prediction of the urban meteorological conditions. We are developing an integrated urban modeling system coupled to the WRF/Noah land surface model as a tool to address urban environmental issues and to study urban-atmospheric interactions. This urban modeling system consists of different methods to parameterize urban land use, a consistent treatment of canopy resistance for both NWP and air-pollution applications, surface biogenic and anthropogenic emissions maps, remote-sensing land-use and characteristics at urban scale, and a companion high-resolution land data assimilation system. It was applied to various metropolitan areas (Houston, Oklahoma City, Hong Kong, Tokyo) and evaluated against urban-scale observations, which showed that representing the urban heat island effects is critical to correctly capture mesoscale and urban-scale wind fields. The predictions produced by the coupled WRF/Noah/urban model with 0.5-km grid spacing were used to drive a computational-fluid-dynamic (CFD)-model-based transport and dispersion model for a case study during the Urban 2000 field experiment conducted in the Salt Lake City. The statistical results indicate that the use of the WRF forecast data in conjunction with the quasi-steady CFD-Urban model has resulted in a significant improvement (by four or five times) in micro-scale transport and dispersion model accuracy over using single sounding data.