Empirical mode decomposition of the atmospheric flows and pollutant transport over real urban morphology

The momentum transport and pollutant dispersion in the atmospheric surface layer (ASL) are governed by a broad spectrum of turbulence structures. Whereas, their contributions have not been explicitly investigated in the context of real urban morphology. This paper aims to elucidate the contributions from different types of eddies in the ASL over a dense city to provide the reference of urban planning, realizing more favorable ventilation and pollutant dispersion. The building-resolved large-eddy simulation dataset of winds and pollutants over the Kowloon downtown, Hong Kong, is decomposed into a few intrinsic mode functions (IMFs) via empirical mode decomposition (EMD). EMD is a data-driven algorithm that has been successfully implemented in many research fields. The results show that four IMFs are generally enough to capture most of the turbulence structures in real urban ASL. In particular, the first two IMFs, which are initiated by individual buildings, capture the small-scale vortex packets that populate within the irregular building clusters. On the other hand, the third and fourth IMFs capture the large-scale motions (LSMs) detached to the ground surface that are highly efficient in transport. They collectively contribute to nearly 40% of vertical momentum transport even with relatively low vertical turbulence kinetic energy (TKE). LSMs are long, streaky structures that mainly consist of streamwise TKE components. It is found that the open areas and regular streets promote the portion of streamwise TKE in LSMs, improving the vertical momentum transport and pollutant dispersion. In addition, these streaky LSMs are found to play a crucial role in pollutant dilution in the near field after the pollutant source, while the small-scale vortex packets are more efficient in transport in the mid-field and far-field.