Geoscience Reference
In-Depth Information
as a boundary condition for atmospheric circulation models, have resulted in highly complex model
structures with multiple soil layers and multiple vegetation layers. In effect, these models aim to predict
the way in which the Penman-Monteith canopy resistance changes with water availability and other
factors such as solar radiation, leaf temperature, carbon dioxide concentration, vapour pressure deficit
and position in the canopy. Some recent examples are the Community Land Model (CLM) (Bonan
et al.
,
2002; Oleson
et al.
, 2008); the UK Met Office MOSES model (Smith
et al.
, 2006; Rooney and Claxton,
2006); and ISBA-TOPMODEL used by MeteoFrance (Vincendon
et al.
2010). Such models have a very
large number of parameters for each of the soil and vegetation layers that may be very difficult to estimate
a priori
(and may, in fact, change over time). Interestingly, the runoff generation components of such
models tend to be rather simple (see, for example, Lohmann
et al.
, 1998c). This is a good example that
what is considered important in a model depends, initially at least, on the perceptions and background
of the modeller.
It is worth noting that the availability of meteorological data may be a problem in applying some
of the more demanding methods, including the Penman-Monteith equation. Net radiation, temperature,
humidity and wind speed data necessitate either an automatic weather station to be installed within the
catchment of interest or, at the very least, a high-quality meteorological station to be nearby. When
this is not the case, some of the simpler methods, even the simple sine curve approach, may still
have value.
At many principal meteorological stations, an evaporation pan may also be available, with measure-
ments of the depth of water lost from a reservoir open to the atmosphere, usually on a daily basis. There
are several different sizes of pans in use, even within the USA. Such measurements can give an index of
the rate of
potential evapotranspiration
at a site but the measured rate does depend on the way in which
the pan is exposed and the nature of the surroundings at the site. In general, such pans evaporate more
water than would be lost from the surrounding surface, even for non-limiting water conditions. Thus, in
general, pan evaporation estimates must be multiplied by an empirical pan coefficient to improve the
estimate for potential evapotranspiration for a particular type of surface. For a US “Class A” pan, for
example, the coefficient is of the order of 0.7. Pan coefficients are tabulated in many hydrological texts
(e.g. Bras, 1990) and the UN FAO have proposed a widely used set of coefficients to adjust measured
pan evaporation to water use by different crops (see the review by Pereira
et al.
, 1999).
3.3.2 Evaporation of Water Intercepted by the Vegetation Canopy
The very low canopy resistance for wet canopy conditions noted above is why, given a source of energy,
the loss of intercepted water from a wet canopy tends to be at higher rates than transpiration from a dry
canopy. The effect is particularly marked for rough forest canopies in windy conditions. This can be taken
account of within the Penman-Monteith formulation of Box 3.1 by allowing high rates of evaporation
at
0 from a conceptual interception storage until that storage is dry, after which transpiration rates
are predicted using dry canopy
r
c
=
r
c
values. The evaporation of intercepted water from leaf surfaces in
rough canopies can be very efficient and is a significant component of the total water balance in some
environments (Calder, 1990). It has also been suggested that there could even be significant losses during
some storm events where the rain falls through unsaturated air such that there is still a humidity deficit
above the canopy. A number of models of the
interception
process have been proposed, of varying
degrees of complexity (Rutter
et al.
, 1975; Gash, 1979; Calder, 1986). The most widely used is probably
the Rutter model which is described in detail in Box 3.2 together with the Calder stochastic model. In
general, without specific measurements of throughfall and stemflow below a vegetation canopy, it is
not possible to identify the parameters of an interception model independently, so that estimating the
parameters of the model depends on finding a study of a similar vegetation type reported in the literature,
although extrapolation from one site to another should be done with care (see Chapter 10).
