Simulated turbulent fluxes over land from general circulation models and reanalyses compared with observations

Turbulent fluxes over land in three general circulation models (GCMs) and the three reanalyses are compared with long-term measurement. Observations were obtained from a variety of sources, including GEBA, Cabauw, The Netherlands, and two newly analysed sites. The intercomparison reveals a wide range of values between the models. The latent heat flux in two of the models shows good agreement with observations. These models benefit from accurately simulated radiative fluxes. The other models tend to overestimate the latent heat flux as long as there is sufficient soil moisture available. This overestimation is caused by a variety of reasons, including: a too intense zonal flow from the ocean to the interior of the continent, which produces an excessive moist advection, thereby increasing precipitation and evaporation; and a high insolation and soil moisture, producing excessive latent heat flux in summer. Where soil moisture is limited too early, an excessive decrease in latent heat flux is apparent towards the end of summer. The sensible heat flux in the models varies when compared with the observations due to the following possibilities: in models that exhibit an excessive Summer drying, too much available energy will be partitioned into the sensible heat flux at the end of summer, since there is no water available for evaporation; in models where radiation is overestimated, the excessive available energy will lead to an overestimated sensible heat flux. Despite the continuous assimilation of observations from the global observing system in the reanalyses, the range of uncertainties is not reduced when compared with the GCMs. Thus, although the reanalysis project was developed to aid the validation of GCMs, it appears to be difficult to accomplish this in the case of the turbulent fluxes. Copyright (C) 2002 Royal Meteorological Society.