Aerosol distributions are of great relevance for air quality especially for cities like Stuttgart with limited air exchange due to its location in a basin. We collected a comprehensive set of data from remote sensing, in-situ methods including radiosondes for the urban background of downtown Stuttgart to determine the impact of boundary layer mixing processes on local air quality and to evaluate the simulation results of the high-resolution large eddy simulation (LES) model PALM-4U at 10 m grid spacing. Stagnant meteorological conditions caused accumulation of aerosols and chemical composition analysis shows that ammonium nitrate (37% ± 9%) and organic aerosol (OA, 34% ± 9%) dominated during this winter study. Case studies show that clouds during previous nights can weaken temperature inversion and accelerate boundary layer mixing after sunrise by up to 3 hours. This is important for ground-level aerosol dilution during morning rush hours. Furthermore, our observations validate results of the LES model PALM-4U in terms of boundary layer heights and aerosol mixing for 48 hours. The simulated aerosol concentrations follow the trend of our observations but are still underestimated by a factor of 4.5 ± 2.1 due to missing secondary aerosol formation processes, uncertainties of emissions and boundary conditions in the model. This paper firstly evaluates the PALM-4U model performance in simulating aerosol spatio-temporal distributions, which can help to improve the LES model and to better understand sources and sinks for air pollution as well as the role of horizontal and vertical transport.