Hourly observations of eddy-covariance water vapour fluxes obtained over a
Douglas-fir forest are used to evaluate three transpiration models. The mod
els are (1) the Penman-Monteith model with a Jarvis type of formulation for
the canopy resistance, including an explicit function for specific humidit
y deficit, (2) the Penman-Monteith model in which the specific humidity def
icit response is replaced by a response to transpiration rate itself as des
cribed by Monteith (1995), and (3) a modified Priestley-Taylor formula. Mod
el parameters are optimised against the observations of 43 dry days during
the growing season. Systematic differences between model and observed trans
piration could be related to wind direction. These deviations correspond wi
th deviations found in the observed energy balance for the same wind direct
ions. The mean square of the residuals is approximately two times larger th
an the value found for the uncertainty in the eddy-covariance measurements
due to atmospheric statistics. Distinct responses to the specific humidity
deficit, solar radiation, soil matrix potential and shoot growth are found.
No temperature response is found. The response of transpiration to an incr
ease of leaf area index during shoot growth suggests that the transpiration
from new shoots is higher as that from older shoots. However, other physio
logical changes at the start of the growing season may play a role as well.
An analysis of residuals shows that the Jarvis model gives good results fo
r all conditions encountered. It is shown that the Monteith model can be fo
rmulated such that it is almost equal to the Jarvis model for this aerodyna
mically rough forest. Despite its simple formulation a modified Priestley-T
aylor formula (including LAI and soil matrix potential response) gives reas
onable results, although at moderate irradiation and high specific humidity
deficit deviations are significant. A comparison with results from another
coniferous forest (Thetford forest) shows that transpiration rates are onl
y slightly larger despite the much larger leaf area index. (C) 2001 Elsevie
r Science B.V. All rights reserved.