The atmospheric boundary layer (ABL) over land exhibits a pronounced diurnal cycle driven by radiative heating and cooling of the Earth’s surface. It alternates between a nocturnal stable boundary layer (SBL) and a daytime convective or unstable boundary layer (CBL), with transitional phases between them. Urban wind and thermal environments are strongly influenced by this cycle, as turbulence interacts with surface roughness, topography, and heterogeneous heat fluxes. Understanding these interactions is critical for the development of reduced-order models that can predict flow behavior at neighborhood scale under different atmospheric conditions. While extensive studies have investigated neutral conditions, see for instance our previous works most urban-resolving LES research still concentrates on neutrally stratified flows. Only a limited number of studies address non-neutral ABLs. For instance, Grylls et al. \cite{Grylls} provided guidelines for LES of stably stratified urban flows, while Zhou et al. \cite{Zhou} examined coherent structures and turbulence in the convective boundary layer over canonical urban areas. Nevertheless, the dynamics at building scale under non-neutral conditions remain poorly understood, and the scarcity of high-fidelity datasets hampers progress in this area. In order to go beyond the state of the art, the present study aims at improving our understanding of urban flow dynamics under varying ABL stability conditions and at the same time to generate a highly resolved spatial and temporal datasets for a complex urban geometry, filling the gap in the availability of high-fidelity databases for non-neutral urban flows. To evaluate the impact of atmospheric stability, different atmospheric stability conditions are explored. Here, this is controlled by the streamwise velocity imposed at z = 100 m and the prescribed surface heat flux. A constant potential temperature profile ($\theta = 300 K$) is imposed below 1000 m, with an inversion layer of constant rate applied above this height.