The same reasoning can be applied to the errors in the covariance calculation, i.e., the computa-
tional errors become significant if the fluctuations in both signals are small with respect to their
means (i.e., x ′y′ / xy is small). However, in the measurement of vertical fluxes, the wind signal
oscillates around zero (i.e., w near zero), thus keeping the ratio large enough to avoid errors in 20-
min flux calculations. Based on results reported by Tanner and Green (1989), there were no
significant error ratios in fluctuations of T′ and w′ and its covariance w′ T ′ .
B. Site description
Theoretically, measurements of surface flux by the eddy-correlation method require an extended
upwind surface with reasonably homogeneous flux sources and sinks that is clear of obstacles
capable of disrupting the horizontal streamlines from parallel to the surface. These conditions were
not met in many of the field measurements reported to date, so the results or empirical formulas
developed by the various investigators varied.
In this experiment, the sonic anemometer was installed to sample the northwest prevailing wind
direction. However, during the experiment, though we did not record the change of wind direction
as a function of time, not only sharp variations in the wind speed but sudden changes in wind
direction were frequently observed. Without being equipped with a cut-off device for wind direc-
tions other than the desired one (i.e., northwest), the sonic signals gathered are bound to be rather
complex; they are composed of effects due not only to variations of wind speed and direction, but to
obstacles causing the upwind flow to be neither homogeneous nor parallel to the surface.
As indicated by Yen (1993), a considerable amount of time and effort was directed to securing a test
site that was logistically as well as economically feasible for micrometeorological measurements.
Even after a concerted effort, however, we still were unable to select a site to meet the requirements
outlined above in A, Experimental Setup. A small (about 8 m 8 m) site located behind CRREL's main
building was finally chosen. The site is elevated on three sides and bordered on one side with a similar
but smaller rectangular cavity. All around the site all sorts of obstacles, such as trees, buildings, a hut,
and plants, offer resistance to air flow and create extra air turbulence that contrasts with a normally
homogeneous field that has a large and smooth upwind fetch. Though we knew from the onset that
the results obtained from this non-ideal site would not provide any universal usage, it did enable the
researcher to gain some experience and insight in microphysical measurements and provided a
basis for comparison with other reported results measured under more or less similar conditions.
Along with the sonic measurement, an accompanying measurement of the air temperature (near
the sonic measurement) at various heights (up to 1.29 m) was also carried out. Its aim was to verify
the reported results that there is a temperature inversion near the surface layer. A Campbell CR7
data logger was programmed to take the eight thermocouple readings at sampling rates of 30 s, 1, or
2 min, with the selection of data averaging time ranging from 1, 10, 20, 30, or 60 min.
VI. RESULTS AND DISCUSSION
Sonic turbulent heat flux measurements commenced in May 1991 and continued for as long as
meteorological conditions permitted (it is not feasible to collect data on very windy days or when it
is raining or snowing). Over the period 19911992, 46 sets of flux data were collected. Due to the
lack of snowfall during the winter, only 10 sets of the data were taken over either a shallow snow
layer, a thin ice layer, or a spotty (or patchy) snow cover.
In early April 1992, due to construction activities occurring near the test site and problems with
the meteorological data-gathering facilities for wind-speed direction and net radiation data, this
series of tests was terminated. The test tower and the other associated measuring devices were
moved to the Geophysical Sciences Branch field test site in front of the Frost Effects Research
Facility in August 1992.