Data from: Characteristics and mechanisms of non-stationary turbulence in a megacity area

The observational data were collected from the 325-m meteorological tower in Beijing, China (39.58°N, 116.22°E) from January 1 to December 31, 2017. The site is surrounded by a typical urban landscape, including dense buildings and roads. The sonic anemometer (Model Windmaster Pro, Gill, UK) was equipped at 140 m to measure wind components and air temperature at a frequency of 10 Hz. According to the vertical structure of the urban boundary layer, the observation at 140 m can reflect the turbulence in the surface layer. After quality control and screening of the observation data, 7167 groups of 30-minute segments were retained. ## Description of the data and file structure 1. Turbulence_statistics includes the proportion of each atmospheric stability, and turbulence statistics as a function of non-stationary level. Columns represent non-stationary levels 1-9, and rows represent atmospheric stability (strongly unstable, weakly unstable, near-neutral, weakly stable, strongly stable conditions) and turbulence statistics (turbulent transport efficiency, momentum flux, TKE) respectively in order. 2. Absolute_flux_contribution includes the absolute-flux contribution of four quadrants. Columns 1-4 represent absolute-flux contribution (Q1-Q4, respectively), and the fifth column represents non-stationary levels. 3. fft_mean_overlap and f include energy spectra and natural frequency. The first layer of cells represents u and w variables, the second layer represents stability (unstable, near-neutral, stable conditions, respectively), and the third layer of cells represents the normalized energy spectrum under non-stationary levels. 4. case.zip includes the original turbulence time series for three cases.