206 T.P. Meyers Agricultural and Forest Meteorology 106 2001 205–214
as a National Oceanic and Atmospheric Administra- tion NOAA contribution to the GEWEX Continental
Intercomparison Project GCIP; see Lawford, 1999. Data collected during the last 4 years have provided a
unique opportunity to investigate the inter-annual vari- ability in summertime water and carbon balances for
a rangeland location. In particular, a drought during the summer of 1998 believed to be associated with
the El Niño event provided an opportunity to quan- tify extreme values for parameters related to the local
water and carbon budgets.
2. Methodology
The turbulent fluxes of water vapor, sensible heat, and CO
2
were measured using the eddy covariance technique. Historically, the use of the eddy covari-
ance method Businger, 1986; Baldocchi et al., 1988 has been constrained to mainly short-term intensive
field campaigns. However, improvements in instru- ment design, ruggedness, and stability over the past
decade now allow for nearly continuous measure- ments of sensible and latent energy fluxes using the
eddy covariance technique Goulden et al., 1996; Grelle and Lindroth, 1996; Moore et al., 1996. With
this technique, the average vertical turbulent eddy fluxes of sensible and latent heat and other scalars are
determined as
w
′
χ
′
= P
n i=
1
w − hwiχ − hχ i n
1 where w is the vertical velocity component of the wind
vector, and χ the scalar of interest e.g., water vapor concentration. Here, the bracketed quantities denote
an average or “mean” that is subtracted from the instantaneous values to obtain the fluctuating compo-
nent. Average vertical turbulent fluxes w
′
χ
′
are com- puted in real time using a digital recursive filter 400 s
time constant for the determination of a running “mean”, which is subtracted from the instantaneous
values to obtain the fluctuations from the mean. An averaging period of 30 min denoted by the overbar
is used and is considered large enough for statistical confidence in the covariance quantity, but is also short
enough to resolve the structure of the diurnal cycle.
Wind vector measurements made at experimental sites that are not perfectly flat can result in non-zero
vertical wind velocities measured from the “vertical” coordinate system of the measurement platform. At
the end of an averaging period, vertical turbulent fluxes perpendicular to the mean horizontal wind
which generally follows the contour of the land surface are obtained by mathematically rotating the
coordinate system of the measurement frame of ref- erence sonic anemometer to obtain a zero mean
vertical and transverse velocity w = v = 0. Details of this procedure are outlined by Businger 1986 and
Baldocchi et al. 1988.
The three components of the wind vector were de- termined with a sonic anemometer model R2, Gill In-
struments, Hampshire, England. The stable long-term operational characteristics of this instrument and its
ability to provide measurements during cold weather and light rain events as well as its low power consump-
tion were important considerations in the selection of this anemometer Yellard et al., 1994. The symmetric
head design of the R2 with its slender support structure produces little flow distortion Grelle and Lindroth,
1994 and is well suited for measurements in relatively flat and open sites with short vegetation. Fast response
water vapor and CO
2
concentration measurements were made with an open-path, fast response infrared
gas analyzer Auble and Meyers, 1992. This sensor has been used extensively for flux measurements in
coastal experiments Crawford et al., 1993, and recent ARM Doran et al., 1992 and BOREAS Baldocchi
and Meyers, 1998 experiments. In a recent evaluation of open- and closed-path sensors for water vapor and
CO
2
concentrations, Leuning and Judd 1996 found that for the measurement of CO
2
, this sensor displayed minimal cross-sensitivity to water vapor see Leuning
and Moncrieff, 1990. The gas analyzer is swapped out every 2 months with a recently calibrated sensor.
Calibrations are done both before and after exposure in the field using CO
2
standards obtained from NOAA’s Climate Monitoring and Diagnostics Laboratory
CMDL and a chilled mirror that has certification by the National Institute of Standards and Technology
NIST. Changes in sensor calibration over a 2-month period for both H
2
O and CO
2
have been observed to be less than 5. The sonic anemometer and the
IRGA were placed about 3 m above ground level. Along with the flux measurements derived from the
eddy covariance method, standard meteorological data Table 1 collected at each site included wind speed
T.P. Meyers Agricultural and Forest Meteorology 106 2001 205–214 207
Table 1 Meterological variables measured at NOAA Energy Flux Monitoring Sites along with model number and manufacturer of instrumentation
used Meteorological variable
Manufacturer Model number
Air temperature and RH Vaisala, Helsinki, Finland
50Y Net radiation
Radiation and Energy Balance Systems REBS, Seattle, WA, USA Q
∗
7 Global radiation
LI-COR, Lincoln, NE, USA LI-200 SB
Precipitation Texas Instruments, Dallas, TX, USA
– Wetness
NOAA, Oak Ridge, TN USA –
Soil heat flux REBS, Seattle, WA, USA
– PAR
LI-COR, Lincoln, NE, USA LI-190 SB
Atmospheric pressure Vaisala, Helsinki, Finland
PTB101B Surface temperature
Everest, Fullerton, CA, USA 4000A
Soil temperature NOAA, Oak Ridge, TN, USA
– Soil moisture
Vitel, Chantilly VA, USA Hydra
and direction, air temperature and relative humidity, precipitation, net radiation, incoming global radiation,
incoming and reflected photosynthetically active ra- diation PAR, barometric pressure, ground heat flux,
surface wetness, and soil temperatures at six depths: 2, 4, 8, 16, 32, and 64 cm. A soil moisture sensor
Vitel was installed before the 1997 summer season. This probe, which measures the dielectric constant of
the soil, water, and air matrix, was placed in the mid- dle of a 10 cm soil layer. The soil moisture was then
determined using the methodology outlined by Wang and Schmugge 1980. The surface temperature was
measured with an infrared temperature sensor. These meteorological sensors Table 1 are sampled every 2 s
with a datalogger and multiplexor CR21X, Campbell Scientific, Logan, UT, and averages are computed ev-
ery 30 min, coincident with the eddy covariance data.
2.1. Data acquisition A laptop computer was configured to perform three
operations simultaneously. The priority task is to re- ceive data from the sonic anemometer, comprising the
components of the wind vector, the speed of sound from which the virtual temperature can be derived and
the digitized H
2
O and CO
2
signals from the IRGA. For its second task, the computer retrieves the stan-
dard meteorological data from the CR21X datalogger every 30 min and appends the data to an existing
file. In background the third task, a terminate and stay resident TSR communications program is used
to retrieve these data from the laptop computer via a cellular phone about once every 2 days. The en-
tire system is powered by a bank of nine 12 VDC deep cycle batteries that are charged by eight solar
panels. The system uses approximately 3 A at 12 V, continuously.
3. Site description
