Comparison between single column model simulations to observations taken at a very stable site

It is well known that numerical weather forecast models have serious limitations to properly represent the lower part of the atmosphere under stable conditions. The very low intensity of turbulence mixing is, tipically, the main problem. To model those situations, a local closure assumption is usually taken, which consists in representing turbulence intensity in terms of local stability. In most cases, this is achieved through the proper determination of the turbulent exchange coefficients and mixing lengths.

In this work we compare simulations from single column formulations to vertical profiles of meteorological data observed at a very stable site, localized in a deforested region on the Amazon rainforest. Different formulations for the mixing length are considered. Results show that, in general, the thermal stratification is well represented by the model and that, in some cases, the specific humidity profiles are also well simulated. The wind profile representation is a more difficult task, specially because wind speeds may be very low at the site near the surface. The difficulties are analysed both in general terms and specifically for each turbulence formulation used.