Turbulent Transport in the Atmospheric Surface Layer - Transport in the Atmosphere-Vegetation-Soil Continuum - page 94

More important in this respect is the time span over which averaging takes place.
As turbulence varies at many time scales ( Figure 3.4 ), the 2 minutes shown in
Figure 3.10 provide only a snapshot, containing just the shorter time scales (the stan-
dard deviation of temperature for the full hour of Figure 3.4 is 0.52 K: the 2-minute
snapshot underestimates it by 11%). Hence, to quantify fully the variance in the tur-
bulent signal, as well as the covariance, one needs to average over a period that cap-
tures all relevant time scales: all scales at which vertical wind and the transported
quantity are correlated. For measurements close to the surface (order of 10 m) this
usually leads to averaging times of 10-30 minutes (see also Vickers et al., 2009 ).
However, for measurements on high towers (e.g., above tall canopies, or to study the
higher parts of the surface layer) averaging periods of 1-2 hours may be needed to
capture all lux-carrying scales (Finnigan et al., 2002 ; Schalkwijk et al., 2010 ).
Usually, eddy-covariance measurements are made to determine surface luxes.
However, the measured lux does not simply originate from a point right below the
sensors, but rather from an area mainly upwind of the sensors. Hence, if the area in
which measurements are made is not homogeneous in surface conditions, one needs
to know what part of the area determines the observed lux (so-called 'footprint')
in order to interpret the observations correctly. The size and shape of the footprint
depends on (Horst and Weil, 1992):

• Height at which the observations are made: When observations are made higher above

the ground, the size of the footprint increases: the instrument can 'look' further.
Stability: The size of the footprint increases with decreasing mixing: under unstable con-

•

ditions the footprint is considerably smaller than under stable conditions.
Roughness of the surface: Turbulent mixing is less intense over smooth surfaces, thus

•

increasing the size of the footprint.

Figure 3.11 shows an impression of the dependence of the shape and size of the
lux footprint on stability. The surface area from which 50% of the lux originates is
approximately 32, 320 and 1300 times z m 2 for the three stabilities shown. Thus if the
instruments are installed at 10 m above the ground, 50% of the lux under unstable
conditions comes from roughly 3200 m 2 . The areas for 90% of the lux are roughly
1000, 7000 and 20,000 times z m 2 , respectively.
The eddy-covariance method is widely used to study surface exchange of heat,
momentum, water vapour and CO 2 at ecosystem scale. In many locations long-term,
multiple-year observations are made. Those cover varying land use types and climate
regions. Those long-term observations are often made under the umbrella of regional
(mostly continental) networks, which in turn are coordinated in the FLUXNET project
(Baldocchi et al., 2001 ; Baldocchi, 2008 ). The coordination enables homogenization of
processing and archiving so that it becomes feasible to study land-atmosphere exchange
processes at multiple sites for multiple years (see also Section 8.4 for an example).
Additional information about the eddy-covariance method can be found in Lee
et al. ( 2004 ).