Improving The Accuracy Of Simplified Urban Canopy Models For Arid Regions Using Site-Specific Prior Information

An overly complex urban climate/energy model would not be of much practical use in a decision support setting where a large number of year-long simulations are often necessary. While detailed modeling of urban momentum/energy exchanges can be attempted via Computational Fluid Dynamics (CFD) or mesoscale modeling, conducting multiple full year simulations in a design optimization or sensitivity analysis context is practically impossible. Furthermore, CFD models often do not consider climate/building energy exchanges. To address these limitations, standalone Urban Canopy Models (UCMs) have been developed, attempting to replace the full-fledged atmospheric representation with a computationally light equivalent. In this study, we investigate the impact of several modifications to a standalone Single-Layer Urban Canopy Model (SLUCM) for an arid region based on the prior availability of site-specific information. The suggested improvements are portable to SLUCM schemes incorporated within mesoscale models. The SLUCM will undergo several improvements and each variant will be evaluated based on its ability to predict UHI intensity and air-conditioning energy demand. Three original SLUCM improvements are covered in the present study. (1) We use actual radiation parameters instead of those generated for idealized geometries. It is shown that, in the absence of this improvement, standard UCMs can underestimate the average urban heat island intensity by 5% if using the Town Energy Balance (TEB) radiation scheme or overestimate it by 7% if using the Square Prism Urban Canopy (SPUC) radiation scheme. (2) We use the results of a prior steady-state RANS (Reynolds-Averaged Numerical Simulation) simulation to replace some of the default morphological parameters of the UCM by the more accurate values derived from the RANS results. It is shown that, in the absence of this improvement, standard UCMs using default empirical relations can overestimate the average urban heat island intensity by up to 22%. (3) Instead of using the empirically calculated default value of the urban canyon wind speed, we estimate it directly from the concomitant rural value using a regression model trained by historical measurements. It is shown that, in the absence of this improvement, standard UCMs can underestimate the average canyon wind velocity by more than 12%, although the other indicators (UHI, cooling demand) are not significantly affected.