Walker Branch Watershed Flux Tower Data 1.0 General Information This data set consists of data collected by the NOAA/ATDD at their Walker Branch watershed flux tower site. UCAR/JOSS has not done any quality control nor changed the data in any fashion. The remainder of this documentation file is from NOAA/ATDD. The original form of this document contained some equations and figures that are not provided here. Contact the contacts to get these items. DOCUMENTATION FILE walkerbranchdocument.doc January 28, 1999 AmeriFlux Site: Oak Ridge, TN 1. TITLE EXPERIMENTAL AND THEORETICAL STUDIES OF THE ANNUAL EXCHANGE OF CARBON DIOXIDE AND ENERGY BY A TEMPERATE FOREST ECOSYSTEM 1.2 ABSTRACT This project addresses DOE/TCP program goals by combining continuous, multi-year field measurements of whole-ecosystem CO2 exchange with testing of process-based, plant physiology and ecosystem carbon cycle models. Our overarching goals are to: (1) use micrometeorological methods to quantify the seasonal and annual exchange of CO2 by a temperate deciduous forest, an ecosystem of global significance; (2) quantitatively relate measured intra- and interannual variations in ecosystem CO2 exchange to variations in environmental factors; (3) test sophisticated process models of photosynthesis, plant respiration and growth, and decomposition with field measurements; (4) extrapolate whole-ecosystem CO2 exchange measurements to the regional scale with a mechanistic ecosystem carbon metabolism model; and (5) predict effects of elevated atmospheric CO2 on short- and long-term carbon storage in temperate forests worldwide. Products of this project will include (1) long-term in situ measurements of carbon exchange by a temperate forest ecosystem; (2) improved understanding of processes regulating whole-ecosystem carbon exchange in temperate forests, and by extension, other terrestrial ecosystems; (3) best available estimates of recent carbon exchange by the global temperate forest biome; and (4) mechanistically based predictions of forest CO2 exchange, as forced by changes in ambient CO2 concentrations and climate. 1.4 Revision date of this Document January 28, 1999 2.0) INVESTIGATOR(S) 2.1) Investigator(s) Name And Title. Dr. Dennis Baldocchi Atmospheric Turbulence and Diffusion Division, NOAA PO Box 2456, Oak Ridge, TN 37831 423-576-1243 fax: 423-576-1327 email: BALDOCCHI@ATDD.NOAA.GOV Co-Investigators: Dr. Jeffery Amthor Environmental Science Division Oak Ridge National Laboratory Oak Ridge, TN Dr. Kell Wilson Atmospheric Turbulence and Diffusion Division, NOAA PO Box 2456, Oak Ridge, TN 37831 fax: 423-576-1327 email: WILSON@ATDD.NOAA.GOV Technical Assistance: Mark Hall, Mark Brewer, Randy White and David Auble Atmospheric Turbulence and Diffusion Division, NOAA PO Box 2456, Oak Ridge, TN 37831 2.2) Contacts (For Data Acquisition and Production Information). Dr. Dennis Baldocchi Atmospheric Turbulence and Diffusion Division, NOAA PO Box 2456, Oak Ridge, TN 37831 423-576-1243 fax: 423-576-1327 email: BALDOCCHI@ATDD.NOAA.GOV 2.3) Requested Form of Acknowledgement. Scalar and energy flux data (e.g. CO2, water vapor, sensible heat and solar energy): Co-authorship if there is extensive use of the data to validate models. Acknowledgement if only few data are used to make a supporting point. Meteorological data: Acknowledgment. Acknowledgement: Field data obtained and prepared by Dennis Baldocchi. Atmospheric Turbulence and Diffusion Division, NOAA PO Box 2456, Oak Ridge, TN 37831 3. INTRODUCTION 3.1) Objective/Purpose. The objective of this research is to measure and model air-surface exchange rates of water vapor, sensible heat and CO2 over a temperate broad-leaved forest and to study the abiotic and biotic factors that control the fluxes of scalars in this landscape. Scalar flux densities were measured with tower-mounted measurement systems. Tower-mounted flux measurment systems were installed above and below a temperate forest canopy. This configuration allowed us to investigate the relative roles of vegetation and the forest floor on the net canopy exchange of mass and energy. We also used the tower-mounted flux measurement system to study temporal patterns (diurnal/seasonal) of mass and energy exchange at a point in the landscape. - 3.2) Summary of Variables. Key measured flux variables were net radiation, quantum, latent heat, sensible heat, soil heat and CO2 flux densities above and below the canopy. Key meteorological and soil variables being measured included wind speed, wind direction, air temperature, relative humidity, soil temperature, CO2 concentration, ozone concentration. The micrometeorological measurements are supported with periodic measurements of photosynthetic capacity, stomatal conductance, leaf area index and pre-dawn water potential. - 3.3) Discussion. We measured eddy flux densities of CO2, water vapor and sensible heat and turbulence statistics above and below a temperate deciduous forest near Oak Ridge, TN. The site was topographically undulating and the forest stand was horizontally homogeneous throughout the area deemed as the flux footprint, a region extending over 1 km upwind. One eddy flux measurement system was mounted at 36.9 m above the ground. This system was mounted on a scaffold tower. The sensors on a boom extended 3 m upwind of the tower, to minimize flow distortion. The azimuth angle of the boom was altered to place the instrument array into the wind. The eddy flux densities are determined by calculating the covariance between vertical velocity and scalar fluctuations (see Baldocchi et al., 1988). Wind velocity and virtual temperature fluctuations were measured with identical three-dimensional sonic anemometers. Our experience has also taught us that it is prudent to employ three-dimensional sonic anemometers in forest meteorology applications. When deploying an anemometer over a forest it is nearly impossible to physically align the vertical velocity sensor normal to the mean wind streamlines; sensor orientation problems typically arise due to sloping terrain and to the practice of extending a long boom upwind from a tower. By deploying a three-dimensional anemometer, we are able to make numerical coordinate rotations to align the vertical velocity measurement normal to the mean wind streamlines. CO2 and water vapor fluctuations were measured with an open-path, infrared absorption gas analyzer, developed at NOAA/ATDD (Auble and Meyers, 1992). Fast response meteorology data were digitized, processed and stored using a microcomputer-controlled system and in-house software. Digitization of sensor signals is performed with hardware on the sonic anemometer. Sensor data are output at 10 Hz. Spectra and co-spectra computations show that these sampling rates are adequate for measuring fluxes above and below forest canopies (Anderson et al., 1986; Baldocchi and Meyers, 1991; Amiro, 1990a). Scalar fluctuations were computed, real-time, using a running mean removal method (McMillen, 1988). Analytical and numerical tests show the the recursive filter time constant of 400 s yielded fluxes similar to those computed with the conventional Reynolds averaging approach. Mass and energy flux covariances will be stored at half-hour intervals on high capacity Bernoulli removable disk media. Instantaneous data was recorded periodically. Proper interpretation of experimental results and model evaluation requires detailed ancillary measurements of many environmental variables. Energy balance components that were measured include the net radiation balance, soil heat flux and canopy heat storage 4.0) THEORY OF MEASUREMENTS 4.1 Micrometeorological Measurement Theory. A client of mass and energy flux information want to know how much material is being transferred across the land/air interface. Due to practical and theoretical circumstances micrometeorologists cannot place their sensors directly at this interface. Instead, they must make measurements several meters above the land surface and rely on the application of theories, which are derived from the conservation equations of mass, momentum and energy to interpret fluxes made several meters above the underlying surface. The equation defining the conservation of mass and energy provides the guiding principles for designing and executing micrometerological experiments over land surfaces. Mathematically, this equation can be derived by considering the mass flow of material in and out of a conceptual cube (u c). By applying Reynolds decomposition to the velocity and scalar variables and then time averaging, this equation is expressed, in tensor notation as: (1 The total time rate of change of a scalar (dc/dt) is a function of its local time rate of change plus the advection of material across the lateral. These terms equal the flux divergence and source/sink strengths due to biology (Sb) and chemical reactions (Sch). The terminology associated with tensor notations suggests the space, xi and velocity, ui variables are incremented from 1 to 3. For the space dimension this corresponds to the longitudinal (x), lateral (y) and vertical (z) dimensions. For velocity, this incrementing corresponds with u, v and w velocity vectors, at are aligned in the x, y and z spatial coordinates. For the simple case of steady state conditions (dc/dt = 0), horizontal homogeniety (no horizontal gradients) and no chemical reactions, this equation reduces to: (2 Integrating this equation with respect to height yields the classic relationship, from micrometeorological theory is generally applied. We obtain a relation that shows that the eddy covariance between vertical velocity and scalar concentration fluctuations (measured at a reference height, h) equals the net flux density of material in and out of the underlying soil and vegetation, or the net ecosystem exchange of CO2 (Ne). (3 When the thermal stratification of the atmosphere is stable or turbulent mixing is weak, material leaving leaves and the soil may not the reference height h. Under such conditions the storage term becomes non-zero, so it must be added to the eddy covariance measurement if we expect to obtain a measure of material flowing into and out of the soil and vegetation. (4 While the storage term is small over short crops, it is an important quantity over forests. With respect to CO2, its value is greatest near sunrise and sunset when there is a transition between respiration and photosynthesis and a break-up of the stable nocturnal boundary layer by the onset of convective turbulence. With respect to the study of pollutants, the interception of a wandering plume can cause the storage term to deviate from zero. Advection effects can occur in complex terrain, where drainage flows can occur and across the border of different vegetation or vegetation and natural features such as rivers and lakes. Consideration of the case of advection is often difficult. For it is not straight forward when and which terms should be neglected. Recently, Lee (1998) evaluated the budget equation for CO2 and arrived at the following equation. (5 In the case of Eq. 5 is a mean vertical velocity, which can be induced by mesoscale circulations or topographical drainage. How can we apply the conservation equation to measure fluxes? In the field, we measure fluxes at a given height above the surface, but we want to know the rate CO2 is taken up by the surface below. The vertical flux density of S will remain unchanged with height if the underlying surface is: 1) homogeneous and extends upwind for a considerable distance (this requirement ensures the development of a surface boundary layer); 2) if scalar concentrations are steady with time; and 3) if no chemical reactions are occurring between the surface and the measurement height. Condition one can be met easily through proper site selection. As a rule of thumb the site should be flat and horizontally homogeneous for a distance between 75 and 100 times the measurement height (Monteith and Unsworth, 1990). Condition two is met often for many scalars. Non-steady conditions are most apt to occur during abrupt transitions between unstable and stable atmospheric thermal stratification, during the passage of a front or from the impaction of a plume from nearby power plants. As we attempt to apply micrometerological conditions to over long, time periods and over non-ideal conditions, we must rely on a comprehensive form of the conservation of mass equation and design our experiment on the basis of the terms that need to be assess. For the work at Walker Branch Watershed, we have found that we need to assess Eq. 5 routinely to obtain defensible fluxes, such as respiration during the winter dormant period and large enough values at night that are consistent with the amount of biomass that is respiring. 4.2 Eddy Covariance Technique. The eddy covariance method is a direct method for measuring flux densities of scalar compounds. The vertical flux density is proportional to the covariance between vertical wind velocity (w) and scalar concentration fluctuations (c). A wide range of turbulent eddies contribute to the turbulent transfer of material. Proper implementation of Eq. 1 requires that we sample across this spectrum of eddies. In frequency domain, eddies contributing to turbulent transfer having periods between 0.5 and 2000s typically contribute to mass and energy exchange (Wesely et al. 1989). Hence, wind and chemical instrumentation must be capable of responding to high frequency fluctuations. And computer-controlled data acquisition systems must sample the instrumentation frequently to avoid aliasing and average the signals over a sufficiently long period to capture all the contributions to the transfer. On applying the covariance relation, it is assumed implicitly that the mean vertical flux density is perpendicular to the streamlines of the mean horizontal wind flow. Consequently, the mean vertical velocity, perpendicular to the streamlines of the mean wind flow, equals zero. In practice, non-zero vertical velocities occur due to instrument mis-alignment, sloping terrain and density fluctuations. These effects must be removed when processing the data, otherwise mean mass flow can introduced a bias error (see Businger, 1986; Baldocchi et al., 1988). Evaluating the accuracy of the eddy correlation method is complicated. Factors contributing to instrument errors include time response of the sensor, signal to noise ratio, sensor separation distance, height of the measurement, and signal attenuation due to path averaging and sampling through a tube. Natural variability is due to non-steady conditions and surface inhomogeneities. Under ideal conditions natural variability exceeds about +/-10%, so it is desirable to design a system with an error approaching this metric. Moore (1986) discusses transfer functions for sensor response time and separation distance. We preformed preliminary calculations of transfer function integrals. Corrections due to sensor time constants and separation are less than a few percent. Hence, we decided not to make transfer function to our flux measurements; our experimental design minimized the need for such corrections since we used an open path infrared gas analyzer and a sonic anemometer. Furthermore, these instruments were placed over a tall rough forest, so small distances in physical displacement have little impact on the measurement of scalar flux densities. The sensors which are used to measure CO2 fluxes measure CO2 density fluctuations, rather than mixing ratio. Application of the density corrections, attributed to Webb et al. (1980) are applied to our measurements. 5.0) EQUIPMENT - 5.1) Instrument Description. Eddy covariance flux measurements are made using a triple-axis Applied Technology sonic anemometer and a infrared absorption spectrometer. The sonic anemometer measured vertical (w) and horizontal (u,v) wind velocity and virtual air temperature (T). This anemometer model provides digital output at a rate of 10 Hz. The infrared absorption spectrometer measures water vapor and CO2 density fluctuations. The sensor responds to frequencies up to 15 Hz, has low noise and high sensitivity (20 mg m-3 volt-1). The sensor is rugged and experiences little drift over several weeks of continuous operation. Soil heat flux density is measured by averaging the output of three soil heat flux plates (REBS model HFT-3, Seattle, WA). They are buried 0.01 m below the surface and were randomly placed within a few meters of the flux system. Soil temperature are measured with two multi-level thermocouple probes. Sensors are spaced logarithmically at 0.02, 0.04, 0.08, 0.16 and 0.32 m below the surface. Three thermocouples were used to measure bole temperatures. Sensors are place about 1 cm into the bole and were azimuthally space across a tree at breast height. Canopy heat storage is calculated by measuring the time rate of change in bole temperature in the tree trunks. Photosynthetically active photon flux density and the net radiation balance are measured above the forest with a quantum sensor (LICOR model LI-190S) and a net radiometer (Swissteco Model S-1 or REBS model 6), respectively. A more detailed experimental design was implented at the forest floor because the solar radiation field below a forest canopy is highly variable (Baldocchi and Collineau, 1994). To account for this variability, measurement of solar radiation components were made using an instrument package that traversed slowly across a prescribed domain; the measurement domain is 30 m long under the temperate deciduous forest. Air temperature and relative humidity are measured with appropriate sensors (Vaisala, model HMP-35A). Wind speed and direction was measured with an propeller wind speed/direction monitor (RM Young model 05701). Static pressure is measured with a Vaisala model PTB101B sensor. It operates on a 600 to 1060 mb range over 2.5 volts. Ancillary meteorological and soil physics data are acquired and logged on a Campbell CR-21x data loggers. Half-hour averages were stored on a computer, to coincide with the flux measurements. CO2 concentration profiles are measured with a LICOR 6262 infrared gas analyzer. Samples were normally drawn at 6 liters per minute from sampling ports located at 36.6, 21.7, 9.10 and 0.75 m. Filters are placed at the inlets and at the inlet to the IRGA. The profile is sampled continuously. Solenoid valves were switched every 30 s, we wait 10 s for the old air to be drawn out of the sample cell and then we sample every second for the next 20 s. Cell pressure and temperature is recorded in addition to analog output. Software and equations, derived by LICOR, are used to convert voltages to concentrations during post processing. The LICOR is calibrated and zeroed every day at midnight using gases of known concentration. Pressure of the air outside the cell was measured with a Setra pressure sensor. It operates over an 800 to 1100 mb range over 5 volts. Tests were performed to relate the pressure measured outside the LICOR to those inside the cell using a cell mock up and field tests. Until March, 1998 the reference side of the IRGA was held to zero with a dry chemical absorbent system which consisted of soda lime and magnesium perchlorate. We now purge the reference cell with dry nitrogen. No detectable change in the zero was noted when we changed systems. We also scrub the dry nitrogen of residual CO2 with soda lime. We also suspected a diurnal drift in our zero, which is consistent with the instrument specifications, so starting in March, 1998 we zero the IRGA every 30 minutes. Another potential sampling error may arise because the field measurements are sampled were under negative pressure (750 to 850 mb) as we draw air through the sample cell, while the calibration span gases are under positive pressure as air if forced through the cell, due to plumbing and solenoid configuration. Starting in March, 1998, we vent excess pressure from the span gases so we are sampling and calibrating the LICOR at the same pressure. Field tests, however, indicate that the concentrations in the sensor do scale linearly with pressure, so we doubt this would improve sensor performance significantly. But it eliminates another source of doubt about sensor performance. For the 1995 Isoprene experiment the upper two levels were place over the canopy and were at 41.5 and 34.0 m, a difference of 7.5 m. The solar radiation system of Detlef Matt, that is part of the ISIS project was altered over the course of this study. The table listed below lists the instruments and products that are available. At present all variables are not part of the AmeriFlux output files from the DOE/TCP project, but these data can be made available. 21XCOM4 History for files: WBRyrday.21X 1994-D299, 1995 D299(95)-D115(96) D115(96)-D202(97) >D202(97) --------------------------------------------------------------------------- Table Table Table Table Year Year Year Year Day Day Day Day Hour min Hour min Hour min Hour min 1 Temperature Temperature Temperature Temperature 2 RH RH RH RH 3 R_net R_net R_net R_net 4 Global Global Global Global 5 Diffuse Diffuse Diffuse Diffuse 6 Parin Parin Q15995 Parin Q15995 Parin Q15995 7 UV in UV in UV in UV in 8 NIP NIP NIP 16440e6 NIP 16440e6 9 SD_net SD_net SD_net SD_net 10 SD_Global SD_Global SD_Global SD_Global 11 SD_diffuse SD_diffuse SD_diffuse SD_diffuse 12 SD_parin SD_parin SD_parin SD_parin 13 SD_uv SD_uv SD_uv SD_uv 14 SD_nip SD_nip SD_nip SD_nip 15 pyranometer pyranometer pyranometer pyranometer (Licor) (Licor)PY13520 (Licor) (licor) PY 13520 16 PAR_rotating PAR_rotating PAR_rotating PAR_rotating shadeband shadeband shadeband shadeband Q15993 Q15993 17 SD_pyran SD_pyran Pyran RSB PY24050 Pyran RSB PY 24050 18 SD_Par_shaded SD_Parshaded SD_pyran SD_pyran 19 max_pyran max_pyran SD_Par_shaded SD_Par_shaded 20 max_par_shaded max_par_shaded SD pyran RSB SD pyran RSB 21 min_pyran min_pyran max_pyran max_pyran 22 min_par_shaded min_par_shaded max_par_shaded max_par_shaded 23 uvb max pyran RSB max pyran RSB 24 uvb tot min_pyran min_pyran 25 SD_uvb min_par_shaded min_par_shaded 26 SD_uvbtot min pyran RSB min pyran RSB 27 uvb uvb 28 uvb tot uvb tot 29 SD_uvb SD_uvb 30 SD_uvbtot SD_uvbtot 31 NIP 16417 32 DIF 73_53 33 GLB 73_4 34 sd NIP 35 sd DIF 36 sd GLB - 5.1.1 Principles of Operation. Sonic Anemometer: Three-dimensional orthogonal wind velocities (u,v and w) and virtual temperature (Tv) were measured with a sonic anemometer (Applied Technology, model SWS-211/3K, Boulder, CO). The pathlength between transducers was 0.15 m. The sensor software corrected for transducer shadowing effects (see Kaimal et al. 1990). Virtual temperature heat flux was converted to sensible heat flux using algorithms described by Kaimal and Gaynor (1991). Infrared Absorption Spectrometer: Water vapor and CO2 concentrations were measured with an open-path infrared absorption spectrometer. Details and performance characteristics of the spectrometer are discussed by Auble and Meyers (1992). In brief, the IR beam was reflected three times between mirrors separated by 0.20m, making an 0.80 m absorption path. The response time of the sensor was less than 0.1 s, sensor noise was less than 300 ´g m-3 and its calibration was steady (it varied +/- 3% during the course of the experiment). The sensor was calibrated periodically with three standard CO2 gases mixed in air, whose accuracy was +/- 1%. Soil Heat Flux Transducer: An encapsulated thermopile yields a voltage output proportional to the temperature difference across the top and bottom surfaces. The device has been calibrated in terms of heat flux through transducer corresponding to the observed temperature difference. Instrument Measurement Geometry. One eddy flux measurement system was placed at 36 m above the ground. The instruments were mounted on a boom that extended 3 m upwind of the tower, to minimize flow distortion. The boom was about 10 m above the mean tree height and was positioned in the constant flux layer. An analysis of lateral and longitudinal sensor separation by Lee and Black (JGR, 1994) show that separation distances up to 1.3 m (for our configuration) yield an error less than 3%. Another eddy flux system was positioned near the floor of the canopy. The instruments were 1.8 m above the ground. This location was in the stem space of the canopy and virtually no foliage was present between the canopy floor and the measurement height (Starting Oct, 1997). Manufacturer of Instrument. Sonic anemometer: Applied Technologies 6395 Gunpark Dr. Unit E Boulder, CO 80301 Soil heat transducer: Radiation Energy Balance Systems (REBS) PO Box 15512 Seattle, WA 98115-0512 Elmer N.J. 08318 Net Radiometer: REBS PO Box 15512 Seattle, WA 98115-0512 Elmer N.J. 08318 Pyranometer and Quantum Sensors LICOR 4421 Superior St Lincoln, NE Data logging system: Campbell Scientific P. O. Box 551, Logan, UT 84321 CO2 analyzer LICOR 4421 Superior St Lincoln, NE Temperature and humidity Vaisala Pressure Senors Static Vaisala Cell pressure Setra 5.2) Calibration. The net radiometer, quantum sensors, pyranometer and soil heat flux plates were calibrated originally by the manufacturer. A net radiometer and quantum sensor were new and were used as a transfer standard, initially. The REBS net radiometer is periodically compared to a reference Swissteco lab standard instrument. The quantum sensor and pyranometer are part of Dr. Detlef MattÆs ISIS, which is a subset of SURFRAD project (http://www.srrb.noaa.gov/surfrad/surfpage.htm). We found some internal documentation from REBS that the calibration of the Q7.0 must be increased by 15.9% to equal that of the Q7.1. Recent re-calibration of the REBS sensor based on tests with new instruments are confirming this difference. SURFRAD has adopted the standards for measurement set by the Baseline Surface Radiation Network (BSRN) which is sponsored by the World Climate Research Program (WCRP) of the World Meteorological Organization (WMO). These are 15 Watts/m2 for broadband solar measurements and 110 Watts/m2 for thermal infrared measurements. To achieve these ambitious goals, the broadband solar instruments are calibrated at the National Renewable Energy Laboratory in Golden, Colorado against standards traceable to the World Radiation Center in Davos, Switzerland. At the Calibration Facility's field site at Table Mountain near Boulder, Colorado, SRRB maintains three reference instruments of each type that the SURFRAD instruments are checked against before and after deployment. Flux and mean concentration CO2 analyzers were calibrated against secondary calibration gases. These gases were referenced to standards prepared by NOAA/CMDL (http://www.cmdl.noaa.gov/ccg/refgases.html) Trace gas standards used for measuring CO2 by the Carbon Cycle Group (CCG) of NOAA CMDL are contained in aluminum cylinders purchased from Scott- Marrin, Riverside, California. The cylinders are treated with a proprietary passivation treatment. CCG uses three different size cylinders but most of our standards are contained in 30 liter (internal volume) cylinders. The cylinders are ordered with brass Ceodeux cylinder valves (CGA590) containing all-metal seats and nickel stems. The cylinders are shipped to CMDL with 1380 kPa (200 psig) of dry, ultrapure air. It is important for the cylinders to be dry (and remain dry) during filling and use. Brass cylinder valves rather than stainless steel, are recommended for all trace gas species measured by CCG. The zero and span of the LICOR infrared gas analyzers were measured every day. The water vapor sensor was calibrated against mixed air samples and referenced to data from a chilled mirror dew point hygrometer. Stability of the water vapor calibration was checked in the field by comparing the instrument sensitivity to the output of a Vaisala relative humidity sensor. The relative humidity sensor was new and calibrated by the manufacturer. We also compared the output of the Vaisala relative humidity sensor against a redundant dew point hygrometer. Both sensors yielded identical humidity measurements. The AmeriFlux standard system visited our site on D253-255, 1997 for comparison. - 5.2.1) Specifications. Calibration factors. Sonic anemometer: supplied by manufacturer. 1.0 m s-1/V with sonic pathlength 0.15 m. Carbon dioxide: about 30 mg m-3 volt-1. Water vapor density fluctuations: varies with vapor density. 2.0 g m-3 volt-1 at 6 C and 3 g m-3 volt-1 at 14 C. Soil heat transducer: about 40 W m-2 mv-1 net radiation: 12 W m-2 mv-1 quantum flux density: 180 ?mol m-2 s-1 mv-1 Pressure: 0.184 mb/mv - 5.2.1.1) Tolerance. Precision or sensitivity estimates: Solar and net radiation: 1 W m-2. Air temperature fluctuations: 0.1 K. Vertical wind velocity fluctuations: 0.01 m s-1. Surface radiative temperature; 0.1 K. Other Calibration Information. Soil heat flux plates were recalibrated after the experiment. Their calibration coefficient was 6% greater than the original factory calibration. The new calibration coefficients were used for data processed after December 1994. We compared the output of our Vaisala humidity sensor against a dew point hygrometer. RH(dew pt)=0.028+1.04 RH(Vaisala), r2=0.995. At this time, we have not adjusted our RH data, but users of the data are free to do so. We compared the output of the ATI sonic anemometer versus computations of virtual temperature. Tvirtual=0.74+1.074 Tsonic CO2 gases were originally referenced to NIST standards. We have depleted those gases and recently purchased standards from Dr. Pieter Tans, CMDL/NOAA lab. Net radiometer calibration against Lab standard Swissteco Radiometer. June 23, 1997 AmeriFlux Calibration. D253-255, 1997. Net radiation, quantum sensors, CO2, and fluxes of CO2, water vapor and heat were compared against the roving standard system. Par Atdd = 8.00 +0.959 PAR ameriflux; r2=0.998 Tair atdd=0.900 + 1.068 Tair ameriflux, r2=0.992 Rnet atdd=4.79 + 0.98 Rnet ameriflux, r2=0.998 (after updating Rn REBS with new Swissteco calibration that is tied to a new Q7.1 sensor) Fc atdd= -1.62 +0.944 Fc ameriflux, r2=0.944 LE atdd= 9.3 +1.23 LE ameriflux, r2=0.75 We are inclined to believe the Ameriflux meteorological instruments for they are new and freshly calibrated. On the other hand we are more inclined to believe the Atdd flux numbers for our open path sensor has minimal correction for transfer functions and no line loss of CO2 and water vapor. 6.0) PROCEDURE - 6.1) Data Acquisition Methods. The eddy correlation flux systems were digitized on the tower using the analog to digital converter of the ATI sonic anemometer. The A/D board was a 12 bit system. Digital signals for the three orthogonal wind velocity components, temperature, humidity, CO2 and ozone were transmitted to a 386 computer in the field lab. In house software (FLUX.EXE) displayed the data real-time on screen for scrutiny, computed fluctuations from means and 30 minute flux covariances. Campbell 21-X data loggers were used to sample environmental variables. These data loggers were connected to another 386 computer via digital line and were interrogated every 30 minutes using Campbell Scientific software (TELCOM.EXE). In the spring of 1997, the data acquistion system was updated, using 2 pentium class computers. Raw data is acquired continuously and is written to a 2 Gig hard drive. - 6.2) Spatial Characteristics. - 6.2.1) Spatial Coverage. Flux footprint calculations were done at our lab. We find that most of the flux sensed by our eddy covariance instrumentation comes from a region within 300 m of the tower. The below canopy measurement of net radiation was performed with sensors on a tram that traversed a 30 m transect under the forest. This design was needed to account for high spatial heterogeneity of light near the floor of a forest. - 6.2.2 Spatial Resolution. Flux footprints have been computed with a two-dimensional Lagrangian model. The dimensions of the flux footprint at the Oak Ridge Site is shown below. - 6.3) Temporal Characteristics. n/a - 6.3.1) Temporal Coverage. Nearly continuous since October 1994. A pilot study was conducted between April 1993 and April 1994. Short and intensive field campaigns were conducted at the site in August, 1984 and July, 1992. - 6.3.2) Temporal Resolution. Flux data were sampled 10 times per second and averaged over 30 minutes. Times reported are the ending times of the averaging period. Fluctuations were computed by substracting a running mean average (determined with a digital recursive filter using a 400 s time constant) from instantaneous values. 7.0) OBSERVATIONS/ SITE CHARACTERISTICS - 7.1) Field Notes. The field site is located on the United States Department of Energy reservation near Oak Ridge, Tennessee (lat. 35o 57' 30"; long. 84o 17' 15"; 365 m above mean sea level). Johnson & Van Hook (1989), Hutchison & Baldocchi (1986), Baldocchi & Harley (1995) and Greco & Baldocchi (1996) describe details on the site and experimental setup. For completeness we present a brief overview of the key site and experimental design features. A. Site Characteristics Micrometeorological measurements of carbon dioxide, water and energy exchange rates were made over a mixed-species, broad-leaved forest, growing in the eastern North American deciduous forest biome. The forest stand consists of oak (Quercus alba L., Q. prinus L.), hickory (Carya ovata (Mill.) K. Koch), maple (Acer rubrum L.), tulip poplar (Liriodendron tulipifera L.) and loblolly pine (Pinus taeda L.). The forest has been growing since agricultural abandonment in 1940. The mean canopy height was about 26 m. The peak leaf area index of the canopy typically occurs by day 140 and reaches about 4.9 (Hutchison et al., 1986). In 1995, the site experienced a hail storm on day 135. That storm shredded many leaves in the vicinity of the research tower and reduced peak leaf area index to about four. The soil is classified as a Fullerton series, Typic Paleudult, otherwise described as an infertile cherty silt-loam. Soil moisture of the top 0.15 m soil and litter layer was measured weekly using the gravimetric method. Pre-dawn leaf water potential was measured periodically to obtain an integrated measure of the soil water potential in the root volume of understory saplings. Using a pressure chamber, water potential was measured on six to eight leaves (these data were obtained at a site 6 km from the Walker Branch field on leaves of Acer rubrum and Liriodendron tulipifera). Leaf nitrogen content is about 1.8% 8.0) DATA DESCRIPTION - 8.1) Table Definition With Comments. '30 min data for AmeriFlux Web Page File names: wbw1995.dat wbw1996.dat wbw1997.dat wbw1998.dat These data are subject to several data filters. Filtout searches for outliers and replaces them with 9999. These periods are associated with rain events for the most part. Outliers are defined by limits set for variables according to variance, skewness and kurtosis thresholds. They differ for the sonic anemometer and infrared spectrometer. Filtturb is a filter that screens the data for limits according to Monin Obukov scaling theory. Mostly it looks for limits on the standard deviation of w. FiltCO2 screens the CO2 flux data for physiological limits. Data variables in WBWyear.DAT files Data Output string DT$, rn$, rnet$, rg$, rgd$, PI$, PDIF$, PFL$, wnd$, u$, us$, TA$, rhov$, vpd$, TS2$, TS4$, TS8$, ts16$, TS32$, TB$, wet$, PMB$, le$, h$, g$, CS$, nee$, FC$, fcadd$, fdr$, co2a$, co2b$, co2c$, co2d$, wmeso$, stdpar$, stdrg$, wc3d$, fc1d$, _ t$, q$, c$, tt$, qq$, cc$, w$, ww$ DAYTIME is day number and ending time (Eastern Standard Time) of run RNET is net radiation (W m-2) RNET is a duplicate net radiation (W m-2) RGLOBAL is global shortwave radiation (W m-2) RGDIFFUSE is diffuse solar radiation: (W m-2) PAR_INCIDENT is total photosynthetic photon flux density (?mol m-2 s-1) PAR_DIFFUSE is diffuse photosynthetic photon flux density (?mol m-2 s-1) PAR_FLOOR is photosynthetic photon flux density at the forest floor (?mol m-2 s-1) WND_DIRECTION is wind direction U_WIND_SPEED is wind speed (m/s) FRICTION_VELOCITY_U* is friction velocity (m/s) TAIR_35 is air temperature at 38.2 m (C) RHOV is absolute humidity at 38.2 m (g m-3) VPD is vapor pressure deficit at 38.2 m (kPa) TSOIL2 is soil temperature at 2 cm (C) TSOIL4 is soil temperature at 4 cm (C) TSOIL8 is soil temperature at 8 cm (C) TSOIL16 is soil temperature at 16 cm (C) TSOIL32 is soil temperature at 32 cm (C) TBOLE is bole temperature. (C) WET is wetness sensor, 1 is dry, 0.5 is moist and 0 is wet PRESSMB is static station pressure (mb) LE is latent heat flux density (W m-2) H is sensible heat flux density (W m-2) G is soil heat flux density (W m-2) CS is canopy heat storage (W m-2) NEE is the net ecosystem CO2 exchange. It is the sum of FC_WPL_3D, CO2Storage and FCDRIFT. It equals net ecosystem CO2 exchange (?mol m-2 s-1) FC_WPL_3D is canopy CO2 flux density, measured via eddy correlation (?mol m-2 s-1). The Webb et al. density correction has been applied CO2STORAGE is the storage flux term of CO2 under the eddy flux system (?mol m-2 s-1) FCDRIFT is the mesoscale drift flux induced by drainage and mesoscale flows. The correction is attributed to a report by Lee (1998). CO2_A is ambient CO2 concentration at 0.75 m (ppm) CO2_B is ambient CO2 concentration at 9.10 m CO2_C is ambient CO2 concentration at 21.7 m (or 34.0 m during the 1995 isoprene study). CO2_D is ambient CO2 concentration at 36.6 m (or 41.5 m during the 1995 isoprene study). WMESO is the mesoscale vertical velocity. It is determined by the difference between the measured vertical velocity (for a thirty minute period) and a mean vertical velocity that is a function of wind speed and wind direction. The later represent vertical motion due to zero drift of the sonic anemometer, mean slope and tower effects. On annual average the mean mesoscale velocity approaches zero. (m s-1) STDPAR is the standard deviation of photosynthetically active radiation (umol m-2 s-1) STDRG is the standard deviation of global radiation (W m-2) WC is the covariance between vertical velocity and CO2 density, after a three dimensional coordinate rotation, but without the Webb density correction (umol m-2 s-1) FC1D is the CO2 flux density without any coordinate rotation, but with the Webb density correction. (umol m-2 s-1) T is the voltage from the temperature of the sonic anemometer (volts) Q is the voltage from humidity of the IRGA (volts) C is the voltage from CO2 of the IRGA (volts) TT is the variance in sonic temperature (volts) QQ is the variance in hygrometer humidity (volts) CC is the variance of CO2 volts W is the mean vertical velocity measured by the sonic anemometer, without coordinate rotation. WW is the variance of vertical velocity (after an online 2 dimensional coordinate rotation) 9.0) DATA MANIPULATIONS - 9.1) Formulas. Water vapor and latent heat flux densities 1) Compute latent heat flux by considering temperature variations in the latent heat of vaporization. WQ = WQ_COVARIANCE * H2OCALIBRATION (g m-2 s-1) WE = WQ / (RHOV * TAIR_K / 216.5) (mb m s-1), intermediate derived quantities LATENT_HEAT = 3149000 - 2370 * TAIR_K; latent heat of vaporization LAMBDA = LAMBDA / 1000 LATENT_HEAT_FLUX= LAMBDA * WQ (W m-2) Sensible heat flux density 1) Correct virtual temperature heat flux from sonic to actual heat flux WTCORR = (WT_COVARIANCE - .32 * TAIR_K * WE / 1000) / (1 + VAPOR_PRESSURE_AIR / 100) 2) Correct the specific heat of dry air for moisture SPECIFIC_HUMIDITY = RHOV / (RHOA * 1000) CP_AIR = 1010 * (1 - SPECIFIC_HUMIDITY) + 4182 * SPECIFIC_HUMIDITY SENSIBLE_HEAT_FLUX = WTCORR * CP_AIR * AIR_DENSITY (W m-2) CO2 flux densities WC = WC_COVARIANCE * CO2CALIBRATION 'mg m-2 s-1 1) Apply Webb et al. corrections for CO2 and latent heat fluxes (WPL) ' CO2CONC = CO2_DENSITY * 8.314 * TAIR_K / (PRESSURE_KPA * 44.01) ' NU = CO2CONC* 44.01 / (28.96 * 1000000) RHOA=AIR_DENSITY SIG = RHOV / (RHOA * 1000) SIGTOT = (RHOV / 1000) / (RHOA + RHOV / 1000) TERM1 = 1.6077 * NU * LATENT_HEAT_FLUX / LAMBDA H=SENSIBLE_HEAT_FLUX RHOC=CO2_DENSITY TERM2 = (1 + SIG * 1.6077) * RHOC * H / (TAIR_K * RHOA * 1005) ;CO2 flux density, with 3 dimensional coordinate rotation and application of Webb Pearman, Leuning density corretion FC_WPL_3D = WC + TERM1 + TERM2 '2/5/98, using term of Lee (1998) 'the mean mesoscale or drift velocity, as distinguished from 'that induced by the sonic orientation æwnew is the mean climatic vertical velocity which is a function of æinstrument orientation and underlying terrain. It is determined by æregressing w versus u and v, non-rotated. The sensor coefficients are recomputed every time we move the sonic anemometer, as when we calibrate the IRGA WNEW = AUN * UN + BVN * VN + CWN æmean mesoscale vertical velocity WMEAN = WN - WNEW æ The advective drift flux term, as derived from Lee fcdrift = wmean * (co2d - co2ave) * rhoa * 44 / 29 fcdrift = fcdrift * 1000 / 44 (micromole m-2 s-1) Webb et al correction for LE LE=LATENT_HEAT_FLUX LATENT_HEAT_FLUX = (1 + SIG * 1.6077) * (LE + SIGTOT * H / (1005 * TK)) Canopy heat storage 1) Define canopy biomass and representative specific heat by weighting the water and cellulose contributions. a) define volume of vegetation as the product of canopy height and its basal area. ' b) heat transfer coefficient is volume x Cp/time for 30 minutes CP=4.175 MJ M^-3 C^-1 for water. CP=2.500 MJ ,-3 C-1 for cellulose c) Computed canopy heat capacity using Gower's biomass data i) bole contribution CG1 = DELTA_BOLE_TEMPERATURE * PLANT_COEF TBOLOLD = TBOL ii) heat storage in the air layer DELTA_AIR_TEMPERATURE = TAVG - TOLD iii) BRANCH AND NEEDLE HEAT STORAGE CG1A = 3.9 * DELTAIR CG2 = DELTAIR * 23.1 '36x1.15x1005/1800s TOLD = TAVG iv) latent heat storage in air layer of canopy DELTA_RHOV = RHOVAVG - RHOVOLD CG3 = 48.8 * DELRHOV '36x2442/1800 RHOVOLD = RHOVAVG CANOPY_HEAT_STORAGE = CG1 + CG1A + CG2 + CG3 - 9.1.1 Derivation Techniques/Algorithms. none provided. - 9.2) Data Processing Sequence. Flux covariances are computed in the field by the data acquisition program. Back at home, calibrations are double and triple checked by comparing old and new calibrations and by comparing the mean response of the scalar flux sensors against independent meteorological instruments. Tests are made for energy balance closure to ensure that the data are of reliable quality. Programs are then run to delete periods when the sensors were off line, off range, being maintained or un-reliable due to rain or instrument malfunction. - 9.2.1 Processing Steps and Data Sets. - 9.2.2 Processing Changes. None to report. - 9.3 Calculations. - 9.3.1 Special Corrections/Adjustments. Eddy fluctuations: CO2_DENSITY = CO2_PPM * AIR_DENSITY * 44 / 29 'mg m-3 ' ' Net radiation: - 9.4) Graphs and Plots. None. 10.0) ERRORS - 10.1) Sources of Error. - 10.2) Quality Assessment. Surface energy balance is tested by comparing measurements of available energy against the sum of latent and sensible heat flux. The ATDD/NOAA flux system has been proven to close the surface energy balance over crops and forests. Daily average energy balance closure is on the order of within 12%. - 10.2.1) Data Validation by Source. - 10.2.2 Confidence Level/Accuracy Judgement. The following are the best estimates of accuracy for a single flux estimate: Net radiation +/- 4 to 7% Soil heat flux +/- 10% Latent heat flux +/- 15 to 20 % or +/-30 W m^2, which ever is larger Sensible heat flux +/- 15 to 20 % or +/-30 W m^2, which ever is larger None of these estimates addresses the variability of flux estimates from site to site. Detection limit of CO2 flux system: 0.025 mg m-2 s-1 The intermittency of turbulence limits the sampling error of turbulent fluxes to 10 to 20%. On top of this we have to deal with measurement errors. Fortunately, lots of statistically averaging reveals stable fluxes and small bias errors (< 12%) on the surface energy fluxes. CO2 concentrations +/- 5 û10 ppm, the zero drift is causing problems. David Earle of University of Nebraska computed that potential error in CO2 based on published specs could be +/- 30 ppm!!! - 10.2.3 Measurement Error for Parameters and Variables. 11.0) NOTES - 11.1) Known Problems With The Data. Nighttime CO2 flux data may have problems due to insufficient mixing and drainage losses out of the control volume. This has been an area of active research by our team. As of March, 1998 we may have a theoretical basis for accounting for drainage of CO2 out of the system. We are applying theory developed by Xuhui Lee of Yale, which is based on the conservation equation. Essentially, we assess a mesoscale induced vertical velocity and multiply it times a mean vertical gradient. The addition of this drift term has eliminated the observation of downward fluxes during the winter when leaves were not present. It has also increased the magnitude of nocturnal respiration to values that are consistent with model calculations. Back in 1996 we attempted a similar scheme since we were observing that mean w was a function of cold air drainage. But we did not subtract the measured vertical velocity from the mesoscale mean. This early attempt yielded corrections that were too big. We feel the CO2 concentration data is chronically high by 5 to 10 ppm, as suggested by a test with the Ameriflux reference system. However, we cannot find a major cause except for diurnal and non systematic drift in the zero. We have done numerous tests on the system to track this bias issue down (our one-one comparison slope is rather good). Note we measure analog output, temperature and pressure and we automatically zero and span at midnight. But field sample pressures are about 850 mb due to sucking, while calibrated gases are injected and measure about 1000mb. It all sounds so standard and simple and straight forward. Its hard to understand why 'elevated' concentrations are occurring. Suspecting my code, I had a long and elaborate dialogue with Jon Welles. I sent the code to Jon Welles for intercomparison and we yield the same values (in fact I derive it from their C code). We have been conscious of ground loops since Boreas days and this is not the problem. The voltage on the LICOR and Campbell panels are identical. I was worried about pressure differences. We had an old LICOR cell. We put a port in the center and did numerous pressure difference tests. They are a few Pa, but nothing note worthy. Zeroing is one issue I am chasing a lot. Most recently we went from chemical to N2 zero. No change! I also zero hourly now. I was observing a very strong temperature effect, but this may be due to periods when the flow from the N2 tank may have been insufficient, so a little CO2 may have remained in the sample cell and of course it would be sensitive to pressure and temperature. One issue that may be worth while is how well we can compute zeros and gains. If I use my standard calibrations and zeros I compute gain factors of 1.03 to 1.06. Licor claims that the gain should be about 1.003. Lately I've tried to deduce my zero by assuming the gain is one and looking at the difference between the calculated and measured span calibration and comparing this with my measured zero. If I do this I notice that the measured and estimated zeros are close but may differ by about 30 mv (eg. 280 vs 310 mv). These minor differences can cause a few ppm bias. If we add the published accuracy of 1% separate sensors can start to diverge by 5 to 10 ppm, which is greater than I had thought. The pressure corrections seem linear, based on inputting known span gases at different pressures. We also routinely recompute the estimated CO2 concentration of the span gas before the calibration is updated. These values agree within the known value +/- 2 ppm typically. So even, if we suspect high daytime CO2 concentrations, itÆs difficult to demonstrate this point on the basis of recomputing the concentration of a know gas. The AmeriFlux comparison identified a 1 C difference in air temperature. We have investigated this problem, and as of May 29, 1998 we have corrected Tair for this bias. We found revised documentation from REBS on correcting the output of a Q7.0 to a Q7.1. Net radiation flux densities have been updated as of June 23, 1998. There was a 15% difference between the old and new methods. This difference was confirmed with a new calibration. - 11.2) Usage Guidance. Prior to application of the advection correction we would caution CAUTION should be exercised when using flux data for several hours surrounding dawn and dusk since these are periods of unsteady conditions. In addition, nighttime data should be closely scrutinized. There are periods when CO2 may be draining out from below the instruments. With use of the advection correction, we caution that schemes are under development. Fluxes may need to be updated as we learn more about the physics of the problem and re-fine our correction procedure. At present we compute new sensor orientation functions every time we move the sonic anemometer. 12.0) REFERENCES Relevant Journal Articles on Studies of CO2 and Water Vapor exchange at Walker Branch Watershed Baldocchi, D.D., B.B. Hicks, and T. P. Meyers. 1988. Measuring biosphere-atmosphere exchanges of biologically related gases with micrometeorological methods. Ecology, 69:1331-1340. Baldocchi, D.D. and P.C. Harley. 1995. Scaling carbon dioxide and water vapor exchange from leaf to canopy in a deciduous forest: model testing and application. Plant, Cell and Environment. 18: 1157-1173. Harley, P.C. and Baldocchi, 1995.Scaling carbon dioxide and water vapor exchange from leaf to canopy in a deciduous forest:leaf level parameterization. Plant, Cell and Environment. 18: 1146-1156. Baldocchi, D.D. and C.Vogel. 1995. A comparative study of water vapor, energy and CO2 flux densities above and below a temperate broadleaf and a boreal pine forest. Tree Physiology. 16: 5-16. Baldocchi, D.D. 1997. Measuring and modeling carbon dioxide and water vapor exchange over a temperate broad-leaved forest during the 1995 summer drought. Plant, Cell and Environment. 20: 1108-1122. Publications Generated From this Project Baldocchi, D.D. 1997. Flux footprints under forest canopies. Boundary Layer Meteorology 85: 273-292. Baldocchi, D.D. 1997. Measuring and modeling carbon dioxide and water vapor exchange over a temperate broad-leaved forest during the 1995 summer drought. Plant, Cell and Environment. 20: 1108-1122. Baldocchi, D.D and T.P. Meyers. 1998. On using eco-physiological, micrometeorological and biogeochemical theory to evaluate carbon dioxide, water vapor and gaseous deposition fluxes over vegetation. Agricultural and Forest Meteorology 90: 1-26 Raupach, M.R., D.D. Baldocchi, H. Bolle, L. Dumenil, W. Eugster, F. Meixner, J. Olejnik, R. Pielke, Sr., J. Tenhunen, R. Valentini. 1998. How is the Atmospheric coupling of land surfaces affected by topography, complexity in landscape patterning and the vegetation mosaic? In: Integrating Hydrology, Ecosystem Dynamics and Biogeochemistry in Complex Landscapes. Dahlem Konferenzen. (submitted) Valentini, R., D. Baldocchi, J. Tenhunen. 1998. Ecological Controls on Land Surface Atmospheric Interactions. In: Integrating Hydrology, Ecosystem Dynamics and Biogeochemistry in Complex Landscapes. Dahlem Konferenzen. (submitted) Amthor, J.S. and D.D. Baldocchi. 1998. Terrestrial Higher-Plant Respiration and Net Primary Production. In: Terrestrial Global Productivity: Past, Present and Future. Eds. J. Roy, B. Saugier and H. Mooney. Academic Press. San Diego. (in press). Baldocchi, D.D. and J.S. Amthor. 1998. Canopy Photosynthesis: History, measurements and models. In: Terrestrial Global Productivity: Past, Present and Future. Eds. J. Roy, B. Saugier and H. Mooney. Academic Press. San Diego. (in press). Wilson, K. and D.D. Baldocchi 1998. Seasonal and interannual variability of energy fluxes over a broadleaved temperate deciduous forest. Agricultural and Forest Meteorology (in preparation). Canadell, J., H. Mooney, D. Baldocchi, J. Berry, J. Ehleringer, C. Field, T. Gower, D. Hollinger, J. Hunt, R. Jackson, G. Shaver, S. Trumbore, R, Valentini, B. Yoder. 1998. Ecosystem metabolism and the global carbon cycle. Bioscience. (in preparation). Malhi, Y., D.D. Baldocchi and P.G. Jarvis. 1998. Carbon sinks in tropical, temperate and boreal forests: three case studies. Plant, Cell and Environment. (in preparation) 15.0) GLOSSARY OF ACRONYMS 16.0 Need to add climate data, wind rose information, soil properties, plant properties, species information and percentages WBW1997.DAT Readme file Version: October 21, 1998 1. The newest file merges met and flux data. 2. We now have a data file that includes almost every time of the year. Missing met or flux data are denoted with 9999 rather than skipping a line. 3. We merge with the corrected radiation data from Detlef Matt's ISIS network. These data are for the Rglobal. 4. CO2 fluxes are presented in terms of the eddy covariance, storage term and the new advective term (after Lee, 1998). Net ecosystem is conceptually equal to the sum of the eddy covariance, storage and advection terms. How to implement the advection term is a topic under intense scrutiny and research. Each term suffers from different measurement errors and technical issues, so the summed flux can have some uncertainty. 5. Individual runs suffer from certain statistical sampling error. If these data are to be used for modeling I suggest computing weekly averaged diurnal patterns. 6. We've tried to correct CO2 concentrations for bias errors. One problem we found was that the introduction of soda lime on the nitrogen zero tank scrubbed CO2 but introduced water vapor into the air stream. We've corrected for this problem and now obtain results of CO2 in modest agreement with the AmeriFlux comparison. We do observe a jump in CO2 concentration when we swapped out LICORs around day 168. That stated accuracy of the CO2 measurements is on the order of 1 to 3%, which is probably true for this situation. I would not use these data to compare with Mauna Loa or the CMDL flask network. They are not precise enough. 7. Thresholds were set on accepting CO2 flux data. In the summer we accept data between 10 and û35 ?mol m-2 s-1. During the autumn the bound are û22 and 13 ?mol m-2 s-1 and during the winter the bounds are û2 and 13 ?mol m-2 s-1 8. In computing the mesoscale vertical velocity (w), we now use a sine transform of w/u versus wind direction instead of a multi-linear regression of w on u and v. This newer approach is symmetric about 0 and 360 degrees, while the former method was not. 9. We have QA/Qced the data by plotting most of the variables by hour using PVWAVE. We have attempted to remedy spurious data by correcting bias errors, flagging them as 9999 or substituting them with a reasonable replacement (and flagging that action).