Laboratory Experiments on Entrainment—a Comparison of Two-Layer Systems with Linearly Stratified Systems

For weather and climate models it is of vital importance to correctly forecast the height of the atmospheric boundary layer as it develops under daytime heating. Recent laboratory experiments (Jonker et al 2004) using the Delft saline convection tank facility, cast significant doubts on the validity of the widely accepted Ri-1-law for the growth-rate of the boundary layer, especially for large Richardson numbers (strong inversions). The experiments contradicted the findings both of Large-Eddy Simulations and of thermal convection tanks (e.g. Deardorff and Willis, 1980). It is important to emphasize that in these experiments the initial state of the boundary layer was a so-called two-layer system: a well mixed layer topped with another neutral layer with lower density; the situation is thus characterized by a well-defined density jump (inversion).

In the present study we investigate besides the two-layer system the classical situation where the initial state is linearly stratified. We make use of the same saline convection tank set-up as previously. Its lateral dimensions are 1m x 1m. The concentration fields of a passive scalar (fluorescent dye) are continuously measured by means of planar Laser Induced Fluorescence (PLIF). In this way we can determine entrainment velocities by two different methods: 1) by tracking the location of the mixed layer height in time, 2) by analysing the evolution of the mixed-layer concentration of the dye (for example, in the case of a 'bottom-up' the concentration will be increased due to the (known) surface flux, but diluted by entrainment). From the data we additionally derive the evolution of the inversion strength, from which (the evolution of) the Richardson number is calculated.

We report the observed entrainment rates for different Richardson numbers and compare the entrainment characteristics of linearly stratified systems with those of two-layer systems.