In this study, we examine representations of the turbulent fluxes under stable conditions. We address two questions: 1) How well do turbulence schemes, independent of land models, perform relative to observations during stable conditions? 2) What mechanisms can explain the performance of the stability schemes?
We isolate the turbulence parameterization from the land models by forcing turbulence schemes using observed surface (e.g., surface temperature) and atmospheric (e.g., air temperature) boundary conditions. This approach of running the turbulence schemes independent of the land models allows us to directly test turbulence representations. We find that all stability schemes tested are unable to recreate a functional relationship with observed fluxes during stable conditions at a number of study sites: Snoqualmie Pass (WA), Shallow Cold Pool experiment (CO), and the Cooperative Atmospheric Surface Exchange Study (KS).
One of the most interesting results is that all schemes fail to capture is the presence of counter-gradient fluxes in the observations. During stable conditions, when the air is warmer than the surface, turbulence schemes are required to simulate a sensible heat flux that warms the surface. This assumption is regularly violated as shown by eddy-covariance observations. We propose a mechanism for this behavior that could be incorporated into new stability parameterizations. Finally, we discuss potential paths for the land modeling community to better represent and test turbulent fluxes.