Validation of a data assimilation technique for an urban wind model

The Quick Urban and Industrial Complex (QUIC) dispersion modeling system has been developed to calculate wind and concentration fi elds in cities with buildings explicitly resolved. As opposed to other models which are either limited to a simpli ed gaussian plume without buildings or are computationally expensive and take weeks to grid and solve. The focus of this paper is a new data assimilation technique that improves QUIC-URB, a fast response three-dimensional (3D) diagnostic urban wind model. The QUIC-URB modeling system discussed in this paper was adapted from a previous version, which initialized the flow fi eld with horizontally uniform velocities based on wind speed and wind direction information obtained from a single measurement upwind of an urban area. Previous urban studies have shown that cities are often subject to large scale spatially varying in flows. To account for this spatial heterogeneity, a simple Quasi-3D Barnes Objective Map Analysis Scheme (a Gaussian weighted averaging technique), which initializes the flow fi eld based on multiple sensors and soundings located around the urban area has been implemented. This wind field is then modi ed by QUIC-URB's empirical building flow parameterizations to model the flow around individual buildings. The fi nal flow field is then obtained by ensuring mass conservation. This work is a validation of this multisensor data assimilation QUIC-URB model. The analysis shows QUIC-URB solutions compared to results of a hybrid Reynolds Averaged Navier-Stokes (RANS) solution of the same urban environment using the commercial Computational Fluid Dynamic (CFD) solver, FLUENT. Nine individual vertical velocity profi les located around the urban area are extracted from the FLUENT data set to simulate soundings around three urban environments consisting of an array of containers and three di fferent sizes of flow altering topographies. These velocity profi les are used as input profi les for QUIC-URB's new initialization scheme. The fi nal wind fi elds from QUIC-URB and FLUENT are qualitatively and quantitatively compared. The initial implementation of this data assimilation technique captures the gross eff ects of nonuniform mean wind fi elds around urban areas well. However, there are def ciencies when ingesting localized ow. The local flow eff ects of buildings and other relatively small geometries are spread out beyond their applicable region when input data is sparse. Limiting these localized eff ects to their applicable regions is an area for future research.