Impact of urban morphology on turbulence and drag in idealized building clusters

Abstract Urban morphology exerts a strong control on flow behavior within the roughness sublayer (RSL), yet the interactions between building configurations and momentum transport are still not well understood. In this study, we use large‐eddy simulations (LES) and machine learning to examine how key morphological features affect turbulence and drag in idealized building clusters. Our LES results show that dispersive momentum flux (DMF) contributes significantly to total momentum transport, with the ratio of DMF to turbulent momentum flux (TMF) ranging from to across most configurations and dropping below in the densest arrays. Likewise, dispersive kinetic energy (DKE) often surpasses of turbulent kinetic energy (TKE) within the urban canopy layer. These results highlight the need to incorporate both turbulent and dispersive contributions in urban turbulence modeling. The bulk drag coefficient () also exhibits strong sensitivity to urban form, driven primarily by pressure differences between windward and leeward surfaces. Among the morphological parameters considered, the street‐canyon aspect ratio emerges as the dominant factor, showing an correlation with . We also show that drag estimates are dependent on the diagnostic formulation used, particularly whether dispersive contributions are included. To enable efficient parameterization of the complex vertical turbulence structure, we develop a U‐Net‐based machine‐learning model capable of reconstructing vertical profiles of kinetic energy (including both TKE and DKE components) from morphological inputs alone. Together, these findings refine our understanding of momentum and energy exchange in urban environments and provide a foundation for improved parameterizations of urban canopy processes in larger‐scale meteorological models.