Forests play an important role in biosphere-atmosphere mass and energy exchanges, given that they occupy a significant portion of the land surface. Characterizing forest-atmosphere transfer processes is thus important to our understanding of global exchanges. When modeling biosphere-atmosphere exchanges an analogy between heat and water vapor transfers is often used to both simplify the modelisation and to take advantage of our large knowledge concerning heat transfer. However, this analogy may not always be true, especially for forest canopies, where the architecture and sub-canopy create a spatial distribution of sources and sinks which may be different for heat and water vapor.

This question was investigated during an intensive measurement campaign conducted from 28th of July to 24th of October 1999 at Le Bray (CarboEurope flux site FR2) a 30 year old maritime pine stand (Pinus pinaster Ait) located in the south-west of France. This forest canopy exhibits three distinctive layers: the upper layer, between 13 and 19 m at the time of the experiment, which consists of a dense vegetation layer of branches and needles; the middle layer, between 1 and 13 m, corresponding to the trunks and dead branches; the lower layer or understorey, which consists of grass (Molinia coerulea (L.) Moench).

The experimental set-up consisted of three levels of turbulent measurements of 3D wind, temperature and humidity. Two levels were fixed during the whole campaign, one at 40 m, the other at 5 m, whilst the third level was initially installed at 9 m (until the 24th of August) in the trunk region, and thereafter at 23 m (4 m above the canopy).

From the beginning of the campaign up until the 12th of September no rainfall occurred. Consequently the water stress increased during this period, as observed via the Bowen ratio at 40 m which increased from 2 to 6. Important rainfall was observed on the 13th of September and subsequently several times until the end of the campaign. Thus this second period corresponded to a decrease in water stress.

Analysis of correlation coefficients between measurement levels for temperature and humidity fluctuations showed that in water-stressed conditions, the humidity fluctuations are strongly coupled (r=0.6-0.7) in the vertical, both above and within the forest canopy, while the correlation of temperature fluctuations quickly drops off with height. In the second period of the campaign the correlation of humidity fluctuations between different levels behaves like the temperature correlation.

These results suggest that under water-stressed conditions, heat and water vapor are not transported by the same turbulent structures, as a consequence of the different spatial distribution of sources and sinks for those two scalars. This is currently being investigated through the spectral analysis of the correlation of temperature and humidity fluctuations between different levels, and their respective flux covariances at each level.