In land surface parameterization (LSP) schemes, the energy balance at the surface determines the calculation of the surface temperature. Since most LSP use Monin-Obukhov theory for surface layer flux calculations, problems may arise in the transition between stable and unstable regimes under certain conditions. Particularly favorable conditions are a low wind speed, a wet ground under dry air, and a large roughness length (for temperature and moisture). The problem is the following. In calm conditions turbulent transfer is virtually nil in the stable regime but can be large in the free convection regime. If a large gradient of specific humidity is present between the surface and the air at the top of the surface layer, the transition between no evaporation and a sizeable evaporation (a few hundred W/m2) can take place with a very small change in surface temperature. The magnitude of the temperature change necessary to induce large changes in the energy budget depends on zot.
For LSP schemes using a prognostic equation for temperature, the problem may manifest itself by an overshoot followed by oscillations around the neutral value. For schemes using iterations, convergence may be difficult to achieve. The purpose of this presentation is to draw attention to the nature of the problem and to propose simple solutions