Evaluation of Turbulence Stability Schemes of Land Models for Stable Conditions

During stable conditions, simulated sensible heat fluxes, Q h , in land models exhibit a wide range of behavior, with some models decoupling the atmosphere and land surface, while others continue turbulent exchange. This behavior is dictated by the representation of stability within the model. We explicitly test the representation of Q h during stable conditions using observed surface temperature, air temperature, and wind speed at the Shallow Cold Pool (SCP) Experiment site in Colorado, USA. This site is characterized by frequent stable conditions and weak‐wind conditions, which are known to be difficult to represent in turbulence schemes. We test the log‐linear, Louis, Holtslag and de Bruin, Beljaar and Holtslag, and Cheng and Brutsaert stability schemes against the observations at the SCP site. We also evaluate the minimum conductance assumption commonly implemented in land models. Simulated Q h biases are generally small, due largely to compensating errors. Parameterizations could represent Q h in strong‐wind conditions and during the transition from a weak‐ to strong‐wind regime. The behavior of simulated Q h during strong winds was controlled by the aerodynamic surface roughness, z 0 . All models, except those that maintain a minimum conductance, were unable to represent turbulence in the weak‐wind regime. We discuss the interacting roles of countergradient fluxes, intermittent turbulence, and decoupled turbulence during this regime. Finally, we list the characteristics that a turbulence parameterization should include and recommend that land models should include a minimum conductance to simulate Q h in the weak‐wind regime.