A model inter-comparison and sensitivity analysis of three urban canopy models

This study summarizes the inter-comparison of three urban canopy models that are able to derive local-scale urban air temperature from rural atmospheric data and urban land use and built form characteristics at an hourly resolution. The focus of this presentation are the findings of the sensitivity analysis and the lessons learned in the process. The evaluated models are the Urban Weather Generator (UWG), the Vertical City Weather Generator (VCWG), the Surface Urban Energy and Water Balance Scheme (SUEWS). The evaluation is done against a two-week-long air temperature and relative humidity measurement conducted in the neighborhood of Újlipótváros in Budapest, Hungary. 
 
The study found a good agreement between modeled and observed air temperature values with a root mean square error (RMSE) remaining between 1–2ºC when calculated for the entire period. However, when separated per day- and nighttime, as well as per cyclonic and anticyclonic periods, the RMSE of the models increased up to 2–3ºC—particularly when calculated for nighttime and/or for anticyclonic periods. The sensitivity analyses shed light on additional shortcomings in the models. It revealed UWG’s low sensitivity to trees and vegetation. In this regard, especially the presence of trees in the urban canopy were not captured by the model. The analysis also found discrepancies regarding VCWG’s model physics, as the model responded to the increase in shortwave radiative forcing with decreasing air temperatures. The study will conclude with a set of recommendations for model developers and users.