230 J.-C. Calvet Agricultural and Forest Meteorology 103 2000 229–247
be calculated according to A
n
, consistent with obser- vations showing the strong correlation between water
use and CO
2
assimilation Cowan, 1982. The net as- similation computed by a coupled physiology-SVAT
model such as ISBA–A–g
s
Calvet et al., 1998 or other models Ji, 1995; Dickinson et al., 1998 can
be used to estimate LAI according to the prescribed climate and CO
2
concentration and, hence, to ex- plore biosphere feedback mechanisms in response
to changes in rainfall patterns, temperature, and soil water storage. Since, in most cases, the hydrology of
the root-zone and the surface fluxes are controlled by vegetation, modelling the rate of soil water extraction
by the plant roots and the stomatal feedback is impor- tant for atmospheric, hydrological, and environmental
studies De Rosnay, 1999. In SVAT models, the effect of soil water stress on plant transpiration is generally
represented by applying a function depending on soil moisture or soil water potential to stomatal conduc-
tance or to the parameters of photosynthesis. The shape of this ‘stress function’ varies a lot from one
SVAT model to another Mahfouf et al., 1996. The rather straightforward assumptions defining the stress
function in present SVAT models may be adequate to represent large scale phenomena. However, more thor-
ough parameterisations may be useful for mesoscale meteorological or hydrological applications, in which
coherent landscape units may be identified.
In this paper, a large number of results obtained at the leaf level, and micrometeorological data at the
canopy level, are analysed in order to better under- stand the intra- and inter-specific variations of the pho-
tosynthesis parameters in both stressed and unstressed conditions. Three models of the stomatal conductance
are employed to discriminate the soil water stress from the atmospheric water stress. In the last section a sim-
ple representation of these effects is implemented into ISBA–A–g
s
and the model is applied to three annual vegetation cycles on the MUREX modelling the us-
able soil reservoir experimentally fallow site Calvet et al., 1999.
2. Modelling g g
g
s
: the Jarvis and the A A
A–g g
g
s
approaches
The exact mechanism for the stomatal humidity response is still unresolved Jacobs, 1994. A number
of mechanisms have been proposed concerning how the effect of saturation deficit should be represented.
For example, Bunce 1985 refers to the direct ef- fect of cuticular evaporation on stomata, while Mon-
teith 1995 uses the interaction between the water vapour flux and the stomatal conductance. Depending
on which mechanism is considered, the humidity de- scriptor may be either the relative humidity at the leaf
surface or the water vapour deficit. In the leaf gas ex- change studies considered here, the variable employed
to characterise the effect of air humidity on leaf con- ductance g
s
was the leaf-to-air saturation deficit D
s
, which may be defined as
D
s
= q
sat
T
s
− q
a
1 where T
s
and q
a
are leaf temperature and air specific humidity, respectively.
Various ways to apply both Jarvis-type and A–g
s
ap- proaches have been proposed in the past. In this study,
two Jarvis-type approaches were employed, together with the A–g
s
model proposed by Jacobs et al. 1996. The different parameterisations of the unstressed g
s
are presented below. 2.1. A Jarvis parameterisation based on surface
temperature inputs Classical examples of the use of Jarvis-type pa-
rameterisations of stomatal conductance are given by Sellers et al. 1986, Noilhan and Planton 1989, and
Shuttleworth 1989. The temperature dependence of the leaf conductance g
s
may be based on either the leaf temperature T
s
or the air temperature T
a
Section 2.2. In optimal soil moisture and leaf temperature condi-
tions T
s
= T
o
, g
s
may be taken to have the form g
s
T
o
= 1 − α
∗ H
D
s
r
∗
smin
1 + 11.1R
g
R
gL
2 where T
o
is the optimal leaf temperature, D
s
the leaf-to-air saturation deficit as defined by Eq. 1,
α
∗ H
the parameter representing the stomatal sensitivity to air humidity, r
∗ smin
the minimum stomatal resis- tance, R
g
the incoming solar radiation, and R
gL
is the limit value of global radiation set to 100 W m
− 2
in this study.
The T
s
-dependence of g
s
is specified as g
s
T
s
= g
s
T
o
T
s
− T
1
T
2
− T
s a
T
o
− T
1
T
2
− T
o a
3
J.-C. Calvet Agricultural and Forest Meteorology 103 2000 229–247 231
with a =
T
2
− T
o
T
o
− T
1
4 In this study, the values of T
1
, T
o
, and T
2
were set to 0, 30, and 40
◦
C, respectively Shuttleworth, 1989. 2.2. A Jarvis parameterisation based on air
temperature inputs In divers SVAT models, the temperature dependence
of g
s
relies on air temperature, rather than on leaf tem- perature. Since, in most cases, T
a
may vary differently from T
s
, it is important to investigate the effect of using T
a
instead of T
s
on the description of soil water stress. For example, in Noilhan and Planton 1989, the value
of g
s
is derived from Eq. 2, where the value of D
s
is replaced by the air saturation deficit D
a
, defined as D
a
= q
sat
T
a
− q
a
5 Also, Eq. 3 is replaced by
g
s
T
a
= g
s
T
o
1 − T
o
− T
a 2
T
o
− T
1 2
6 with T
1
=
◦
C and T
o
= 25
◦
C Noilhan and Planton, 1989.
Since the leaf gas exchange studies considered here are based on the use of T
s
, the use of T
a
will be inves- tigated in the case of the micrometeorological datasets
only. 2.3. An A–g
s
parameterisation In the A–g
s
physiological module of the ISBA–A–g
s
model Jacobs et al., 1996; Calvet et al., 1998, the parameters governing the magnitude of g
s
and its sen- sitivity to leaf-to-air saturation deficit D
s
are, respec- tively, the unstressed mesophyll conductance g
∗ m
, and the maximum leaf-to-air saturation deficit D
∗ max
Ap- pendix A. The parameter g
∗ m
conditions the maximum attainable stomatal conductance, while D
∗ max
represent the sensitivity of stomatal aperture to air humidity.
Typical values of the parameters of the A–g
s
model, for either C
3
or C
4
plants, are displayed in Table 1. This table does not include values of unstressed g
∗ m
and D
∗ max
which are believed to display more variability
Table 1 Standard values of the parameters of the A–g
s
model according to the plant type C
3
or C
4
adapted from Jacobs et al. 1996
a
Plant type Parameter X X
25
Q
10
T
1 ◦
C T
2 ◦
C C
3
ε mg J
− 1
0.017 – –
– f
0.95 –
– –
Ŵ ppm 45
1.5 –
– g
m
mm s
− 1
– 2.0
5 36
A
m,max
mg m
− 2
s
− 1
2.2 2.0
8 38
C
4
ε mg J
− 1
0.014 – –
– f
0.60 –
– –
Ŵ ppm 2.8
1.5 –
– g
m
mm s
− 1
– 2.0
13 36
A
m,max
mg m
− 2
s
− 1
1.7 2.0
13 38
a
ε is the maximum quantum use efficiency, f
the maximum potential value of the ratio between internal and external leaf con-
centration of CO
2
, Ŵ the compensation point, g
m
the mesophyll conductance, and A
m,max
the maximum net assimilation of the leaf. The Q
10
, T
1
and T
2
values modulate the sensitivity of each pa- rameter to temperature through either XT
s
= X
25
Q
T
s
− 2510
10
or XT
s
= X
25
Q
T
s
− 2510
10
{[1 + exp{0.3T
1
− T
s
}][1 + exp{0.3T
s
− T
2
}]}, where XT
s
and X
25
are the values of the parameters cor- responding to the leaf temperatures T
s
and 25
◦
C, respectively.
between plant species. Also, the cuticular conduc- tance g
c
, allowing diffusion of water vapour and CO
2
through leaf cuticle, is accounted for Appendix A. In this study, the following average values of g
c
were employed: 0.25 mm s
− 1
for herbaceous C
3
plants, 0.17 mm s
− 1
for woody plants other than conifers, 0.05 mm s
− 1
for conifers, and 0.15 mm s
− 1
for C
4
plants. They were derived from the review study of Kerstiens 1996.
The leaf temperature dependence of g
s
is accounted for by Q
10
functions applied to the parameters of the photosynthesis model Table 1. This is different from
the Jarvis-type approaches presented before, in which the temperature response is applied to the value of g
s
.
3. Datasets
