The dissimilarity between the turbulent transport of momentum and heat under unstable conditions and its physical mechanisms are investigated in this study, based on the multiple-level turbulence observation from Tianjin's 255-m meteorological tower. The transport dissimilarity is observed from the surface layer to the lower part of the mixed layer as atmospheric instability increases. Although the transport dissimilarity is accompanied by the development of plumes and thermals under unstable conditions, plumes and thermals can produce intense transport of momentum and heat simultaneously. It is convective circulations related to vigorous thermals that cause transport dissimilarity. The horizontal divergence induced by convective circulations imposes a dominant large-scale reduction in the along-wind velocity component near the surface, while the temporal variations in temperature mainly reflect the role of plumes and thermals. This difference in respective physical processes subsequently leads to dissimilar transport between momentum and heat under unstable conditions. Therefore, the influence of convective circulations on the momentum-flux estimation should be considered in atmospheric numerical models, particularly for the simulation and prediction of severe convective weather.

The observational data were collected from the Tianjin 255-m meteorological tower (39.08°N, 117.21°E) from July 1 to August 31, 2017. The meteorological tower is located in the PBL Meteorological Observation Station, Tianjin Meteorological Administration. The station is surrounded by buildings, roads, and vegetation, and is well representative of the typical urban landscape with flat terrain. Five sonic anemometers (CSAT-3, Campbell Scientific, Inc., USA) were equipped on the meteorological tower at the height of 40 m, 80 m, 120 m, 160 m, and 200 m respectively, to measure wind components and air temperature at a frequency of 10 Hz.

The raw data from sonic anemometers are processed over 30-min intervals using EddyPro v7.0.6 (LI-COR Inc., USA) software, to obtain Reynolds means and fluctuations. Then, turbulence statistics, such as sensible heat flux, momentum flux, friction velocity, variances, as well as turbulent transport efficiency for heat and momentum, are calculated using the eddy covariance method. For a given 30 min interval, the flux contribution and the time fraction of different turbulent motions are quantified using the quadrant analysis. The energy spectra, cospectra, and phase spectra are calculated using Fast Fourier technique.