Spatializing the roughness length of heterogeneous urban underlying surfaces to improve the WRF simulation-part 1: A review of morphological methods and model evaluation

Abstract

Roughness length (Z0) is a critical parameter in atmospheric modeling that affects the land–atmosphere interactions. However, atmospheric models often use a fixed and relatively lower value for Z0 in urban areas, which cannot faithfully represent the actual heterogeneous urban underlying surfaces and reproduce the meteorological conditions. In this study, we provided a review of eight semiempirical models based on morphological methods from the literature and adopted them to develop a dataset of nonuniformly distributed roughness lengths in urban areas of Guangzhou, a highly urbanized city in China. In addition, we employed these datasets in the WRF model and evaluated the model. Overall, the estimated Z0 values in dense high-rise areas of the city were significantly higher than those in other regions. Different semiempirical models were found to generate significantly different Z0 values. The average local Z0 values of most models (7 out of 8) were higher than the model default value (0.5 m). Substituting the model default Z0 values with the localized non-uniform Z0 map revealed that the improvement in the simulated temperature and relative humidity at 2 m height was relatively small, whereas the improvement was significant for the simulated wind speed at 10 m height (WS10). The WRF simulations using the C71, T93, and K13 (see the main text for their definitions) schemes to estimate Z0 demonstrated the best improvements in the WS10 simulation of 54%, 50%, and 39%, respectively, compared with that using the default value. This improvement is of great value because many atmospheric models overpredict WS10. The simulation improved not only for the urban sites but also for the non-urban sites, demonstrating an improved ability for all stations. Among the schemes, the C71 and T93 semiempirical models were only applicable for low-density cities, whereas the K13 model applied to cities with significant urbanization. This study revealed the benefit of spatializing the heterogeneous urban roughness length to improve the meteorological simulations. This modification can be employed in the air quality model and further improve the simulations and forecasts of pollutant concentrations.