Study of 'soft' night-time surface-layer decoupling over forest canopies in a land-surface model

The cause of a night-time land-surface model cold bias over forest canopies at three different sites is studied in connection with various formulations of turbulent transfer and the phenomenon of decoupling between the surface and the boundary layer. The model is the Canadian Land Surface Scheme (CLASS), a leading internationally known model that has been tested over a variety of instrumented sites. The bias was first attributed to a deficient turbulent transfer and a few formulations were compared. One formulation is the classical log-linear profile with a sharp cut-off of the fluxes at a critical Richardson number around 0.2, while in the other ones the flux decreases less rapidly with increasing static stability. While the surface-layer formulations have an impact on the modelled canopy temperature, other causes were found for the negative bias. The CLASS model neglected the heat capacity of the air trapped inside the canopy and its inclusion multiplied the effective heat capacity of the canopy, by a factor ranging from 2.3 to 3.4 for the canopies studied, and reduced the error. A correction was also made to the air specific humidity at canopy level and the topsoil thermal conductivity was changed from that of organic matter to that of mineral soil. With these modifications, and using the incoming longwave radiative flux instead of the net longwave flux, the bias almost completely disappeared. Using a scheme with more heat transfer at large static stability, obtained by assuming that the fluxes decrease in magnitude with height in the surface layer, reduced the original bias while using the log-linear formulation amplified the cold bias. The impact of the turbulent transfer formulations is much reduced when they are applied to model runs in which the other above modifications have been made. The phenomenon of decoupling is presented and its understanding is complemented with the new notions of `hard' versus `soft' decoupling and complete versus incomplete decoupling, depending on the impact decoupling has on the model and on the effectiveness of the model in achieving the decoupling. The geostrophic wind speed is a determining factor in separating cases of hard decoupling (rare) from the soft cases (frequent) while the completeness of the decoupling primarily depends on the form of the turbulent transfer curve as a function of static stability.