Role of the shear and inversion strength during sunset turbulence over land: characteristic length scales

The decay of convection induced turbulence over land at sunset (“sunset turbulence”) is investigated by using large-eddy simulations (LES). The study focuses on the role played by shear and the potential temperature jump at the inversion during such situation.

Four atmospheric boundary layers characterized by decaying turbulence are studied by LES using fine resolution and large integration times. The numerical experiments prescribe initially the same surface heat flux, boundary layer depth and potential temperature in the mixed layer. The differences between the simulations are based on the initial potential temperature jump at the inversion (1 K and 5 K) and on the presence or absence of shear (i. e. the presence or absence of a geostrophic wind, which is taken constant with height). The decay process is started by setting the surface sensible heat flux to zero (=sunset), after two hours, which approximately corresponds to 10 turnover times, h/w_m.

In all the simulations, while temperature fluctuations decrease immediately after sunset, turbulent kinetic energy (TKE) remains constant during one eddy turnover time. Thereafter, the decay of turbulence critically depends on the presence of shear and on the magnitude of the inversion temperature jump. Before tw_m/h=3, the presence of shear leads to a lower decay rate for the TKE. In these simulations, the vertical velocity fluctuations decay much faster than the horizontal velocity fluctuations, increasing the anisotropy. In general we find this scenario to continue, implying the anisotropy to increase rather than to decrease, i. e., for large times, tw_m/h>10, turbulence does not relax to an isotropic state.

The “demixing process” (return to equilibrium levels of warm air parcels) discussed by Nieuwstadt and Brost (1986) is reproduced in the simulations, but remarkably only for the particular atmospheric boundary layer characterized by a small inversion jump and no shear. The demixing process can be characterized by observing the buoyancy flux in the boundary layer (BL). In all the simulations, the flux at sunset becomes negative in the BL after tw_m/h=1. In the particular case where demixing process happens, the flux clearly turns to positive for 3<tw_m/h<4 approximately. In the other studied cases the buoyancy flux remains negative in the BL during sunset.

In the process of the sunset turbulence, the characteristic length scales evolve with time. By determining the peaks in the variance spectra of the LES fields, we calculated the characteristic length scales for wind and potential temperature. For the two shear cases, the characteristic length scale for potential temperature increases immediately after sunset and becomes approximately equal to the horizontal domain size. On the contrary, the characteristic length scale for the vertical velocity remains of the order of the boundary layer height during the simulation. For the cases without shear, the rise in the characteristic length scale of the potential temperature happens for tw_m/h>1. From this moment, differences exist depending on the jump of the temperature at the inversion. For 5 K, the evolution of the characteristic length scale for the vertical velocity and potential temperature is comparable to the shear cases. The particular case with small temperature inversion jump (1 K) presents a similar evolution, except during the demixing process (positive buoyancy flux in the BL after tw_m/h=1): in this period, even the characteristic length scale of vertical velocity suddenly increases, becoming of the order of the horizontal domain size for a while. This hints at a subtle relation between the demixing process and the characteristic length scales.

Nieuwstadt, F. T. M. and R. A. Brost: The Decay of Convective Turbulence. J. Atmos. Sci., 43, 532-546. 1986.