Mesoscale urban boundary layer simulation via multi-layer urban parameterization

Abstract

The mesoscale Weather Research and Forecasting (WRF) model with the Building Effect Parameterization (BEP) multi-layer module has been widely used by the research community in recent years to study the effect of the urban area on urban climate, air quality and boundary layer dynamics. However, very few studies have been using this sophisticated multi-layer approach in the Pearl River Delta Region due to the lack of detailed urban morphological information. The impact of different urban canopy parameters (UCPs) on the boundary-layer structure is not documented well, which is important to understand the urban boundary layer better and provide implications for model improvements. In this thesis, parametric studies were carried out through idealized simulations to better quantitatively understand the relative importance of different UCPs on the atmospheric boundary-layer structure and the model’s sensitivity. Through these studies, the inflection point of the vertical wind profile and the well-known urban heat island are well reproduced by the model in highly urbanized region with tall-building clusters. The relative importance of the common UCPs on the boundary-layer is also investigated. It was found that the building height, coverage of buildings are similarly important, whereas the detailed clustering of buildings in the urban grids are relatively less crucial. Therefore, we developed methods to retrieve the urban impervious surface fraction and the building footprint. An urban fraction database was created from the recent high-resolution (10m) Sentinel-2 satellite images and its accuracy was found to be better than the widely used World Urban Database and Access Portal Tools (WUDAPT) level-0 dataset when compared to the Google Earth satellite images. A novel Matlab algorithm was also developed to retrieve a realistic building dataset for Hong Kong. This method provides a solid foundation to improve the accuracy of urban boundary layer simulation for major cities in the PRD region in the future. However, since the detailed urban morphology database might not be available in many countries at this point of time. The WUDAPT level-0 data, which provides a first-order approximation of the building information by the classification of local climate zone through machine learning methods, is also used to drive the multi-layer BEP model in Hong Kong and compared with the one driven by realistic 3D building database to evaluate the model’s performance in a highly urbanized region with sparse building morphological information. Insight and guidance are given to optimize the implementation of the WUDAPT dataset. A real case study was also carried out to assess the feasibility of utilizing the multilayer BEP model to improve the current site-wind availability dataset from the Planning Department of Hong Kong. Monthly time-series comparison with urban stations show significant improvement from the multi-layer BEP model compared to the traditional bulk scheme when compared to urban observation stations. The analysis of vertical wind profiles reveals that the BEP model is simulating the effect on urban heterogeneity at different locations which is not well-resolved in the bulk scheme, making it a much more attractive tool for air ventilation studies.

Date of Award	2020
Original language	English
Awarding Institution	The Hong Kong University of Science and Technology