HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (7) Citing Articles via Google Scholar Google Scholar Articles by Ham, J. M. Articles by Heilman, J. L. Search for Related Content PubMed Articles by Ham, J. M. Articles by Heilman, J. L. Agricola Articles by Ham, J. M. Articles by Heilman, J. L. Related Collections Agroclimatology

SYMPOSIUM PAPERS

Experimental Test of Density and Energy-Balance Corrections on Carbon Dioxide Flux as Measured Using Open-Path Eddy Covariance

^a Dep. of Agron., Kansas State Univ., Manhattan, KS 66506
^b Dep. of Soil and Crop Sci., Texas A&M Univ., College Station, TX 77843

^* Corresponding author (jayham{at}ksu.edu).

Received for publication August 24, 2002.

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Eddy covariance is the most direct technique for measuring water, C, and energy fluxes above crops and managed ecosystems. When using open-path gas analyzers, corrections for air density fluctuations due to heat and water vapor flux must be applied, and these corrections are often larger in magnitude than the CO₂ fluxes. Lack of energy balance closure, a common problem when using eddy covariance, implies that CO₂ fluxes often are underestimated. Research was conducted to evaluate performance of the density corrections by making eddy covariance measurements above a large parking lot where CO₂ and water vapor fluxes were almost zero. Uncorrected and corrected flux measurements were compared to the "known" values to determine accuracy. Data also were collected from a tallgrass prairie and a cedar forest to examine how density corrections and adjustments for energy balance closure affected daily C balances. Raw measurements from the parking lot showed apparent, density-induced, downward CO₂ fluxes (i.e., apparent photosynthesis) of approximately-0.4 mg m^-2 s^-1 that were correlated with sensible heat. On average, the daily uncorrected CO₂ flux was-12.7 g m^-2 d^-1, but the density correction reduced and changed the direction of the flux to 1.8 g m^-2 d^-1, which was very close to independent chamber measurements of 2.8 g m^-2 d^-1. Density corrections in the forest and prairie changed average daily CO₂ fluxes by 20 to 80%. Energy balance closure averaged 80 and 95% in the prairie and forest, respectively. Corrections based on energy balance closure changed daily C balances by 16 to 35%. A plethora of post-measurement corrections, coupled with lack of energy balance closure, signals the need for additional research before eddy covariance can be routinely applied in agronomic research.

Abbreviations: B, Bowen ratio(s) • DOY, day of year • G, soil heat flux • H, sensible heat flux • Rn, net radiation • E, latent heat flux

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
THE MOVEMENT OF ENERGY, water, and C between the field surface and the atmosphere is one of the most fundamental processes in the soil–plant–atmosphere continuum. Crop production, in essence, is the act of mining C from the atmosphere and refining and packaging CO₂ molecules into a valuable commodity (i.e., grain or forage). As pointed out by Campbell and Norman (1998), turbulent mixing and transport of air is essential for replenishing the supply of C and removing water vapor surrounding the crop canopy. Turbulent transport in the surface boundary layer also affects sensible (H) and latent (E) heat fluxes, which along with the radiation balance, govern evapotranspiration and canopy temperature. Measuring the movement of energy and mass in the surface boundary layer is one of the most essential tasks associated with micrometeorology. Unfortunately, there are gaps and unresolved issues surrounding these most fundamental of measurements. In this paper, we examine the performance of open-path eddy covariance, a flux measurement technique that was first used in the 1970s (e.g., Desjardins and Lemon, 1974). It saw limited use at first because of technical problems—Desjardins and Lemon used a propeller anemometer to sense vertical wind speed. However, as new sonic anemometers and improved gas analyzers became available, the popularity of the technique increased rapidly. However, like many measurements, eddy covariance can be fraught with pitfalls, and data often require extensive postprocessing to remove sources of error. In this paper, we examine some of the most important corrections that must be applied to flux data obtained using eddy covariance.

Eddy covariance is the most direct technique for measuring heat and mass fluxes between the surface and the atmosphere (Baldocchi et al., 1988). Historically, the complexity and cost of eddy covariance instrumentation has limited its use. However, as sonic anemometers and open-path gas analyzers become less expensive and more reliable, it is probable that researchers from a wide range of disciplines will use this technique. Eddy covariance instruments are well suited for continuous long-term monitoring of field-scale processes. Agronomists, range scientists, and others conducting research on managed systems may select the technique to evaluate how management decisions and interannual variations in climate affect field-scale water and C balances. Furthermore, as new types of gas analyzers are developed, it will be possible to estimate the flux of many different trace gases (e.g., N₂O), isotopes, aerosols, and organic compounds. Thus, eddy covariance could become a useful tool in examining agriculture's impact on air and water quality.

Eddy covariance is often touted as the best available method for flux measurement because it does not require simplifying assumptions about the physics of the boundary layer, unlike flux profile or Bowen ratio (B) approaches. Unfortunately, raw eddy covariance measurements must be corrected to account for a host of factors, many of which result from limitations in the instruments or non-ideal boundary-layer conditions (i.e., advection, non-simple terrain). Corrections include adjustment for density variations caused by fluxes of heat and water vapor (Webb et al., 1980), adjustment to account for the separation distance between the anemometer and gas analyzer (Moore, 1986), adjustments to account for frequency response of the sensors (e.g., Horst, 1999; Massman, 2000; Massman and Lee, 2002), coordinate rotation to account for the slope of the terrain or transducer misalignment (Wilczak et al., 2001), corrections for advection (Paw U et al., 2000), and the removal of long-term trends. When an open-path gas analyzer is used on relatively flat, uniform terrain, the adjustment for density fluctuations is the largest of these corrections and can change the estimate of flux by 20 to 80% (Leuning and Judd, 1996).

Webb et al. (1980) described the principles underlying the density effects and demonstrated how raw CO₂ and water vapor flux measurements can be adjusted. The corrections originate because open-path gas analyzers do not measure CO₂ and water vapor concentrations as mixing ratios (kg kg^-1) but instead measure density (mg m^-3). For the typical daytime case of upward H and E, the air density of the up-moving eddies is less than that of those moving down because up-moving eddies are warmer and contain more water vapor. If one assumes there is no mean vertical flow of dry air (i.e., no source or sink of air at the surface and no terrain effects), there must be a small net upward wind speed to maintain conservation of mass. Webb et al. (1980) argue that the contribution from this small upward velocity component is missed when flux is computed solely from the covariance. Thus, a correction must be added to covariance estimates of CO₂ flux that has the same sign as the water vapor and sensible heat flux. Leuning et al. (1982) evaluated the density correction over a plowed field where CO₂ flux was assumed to be negligible. Kramm et al. (1995) describe a variation of the density correction that uses a weighting method while Paw U et al. (2000) propose corrections that consider the simultaneous effect of both density and advection.

Lack of energy balance closure is another issue that can reduce confidence in eddy covariance measurements. The conservation of energy, represented by the surface energy balance, is the theoretical linchpin micrometeorologists use to test the accuracy of flux measurements and is expressed as

[1]

where Rn is net radiation and G is soil heat flux and the rate change in canopy heat storage, all in W m^-2. In agricultural systems, canopy storage is small, and G is equated to soil heat flux. Available energy (Rn - G), measured by radiometers and soil instruments, must equal H + E measured by eddy covariance; (H + E)/(Rn - G) should equal unity. Eddy flux measurements across a wide range of vegetation types have shown that energy balance closure typically ranges from 0.7 to near perfect closure, with overall averages near 0.82 (Aubinet et al., 2000; Twine et al., 2000). These results suggest that eddy covariance may be underestimating H and E and imply that CO₂ flux might be underestimated as well. Some have proposed methods to adjust E and CO₂ fluxes based on forced closure of the energy balance (e.g., Twine et al., 2000). However, energy balance closure methods have legitimate weaknesses and are forced to make assumptions that could result in errors as large as those they are attempting to amend.

The objective of this study was to examine the accuracy of the density corrections and energy balance closure adjustments on the measurement of CO₂ and water vapor flux when using open-path eddy covariance. Data were collected above a large parking lot where actual CO₂ and water vapor fluxes were essentially zero, but large H from the dry surface created apparent fluxes of CO₂ and water vapor. The density correction was evaluated by quantifying how well the Webb et al. (1980) correction re-zeroed the CO₂ and water vapor fluxes. Flux deviation from zero (i.e., measurement error) also gave an indication of the overall noise in the eddy flux instrumentation. Additional measurements from tallgrass prairie and cedar forest were used to evaluate the effect of the density correction and energy balance closure on calculations of the daily ecosystem C balance.

	METHODS

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
A portion of the research was conducted in July 2002 on a parking lot located adjacent to the Kansas State University football stadium in Manhattan, KS. The parking lot provided a uniform, dry surface with little or no latent heat flux and very low CO₂ fluxes. A sonic anemometer (CSAT3, Campbell Sci., Logan, UT) and an open-path gas analyzer (LI-7500, LI-COR, Lincoln, NE) were deployed 3 m above the surface near the north edge of the lot. Sensor separation was 0.12 m, and the LI-7500 was tipped 15 degrees to the north to minimize any radiation effects on the gas analyzer. During a portion of the study, a 0.013-mm-diam. fine-wire thermocouple was positioned next to the sonic anemometer. Fetch to the south, the prevailing wind direction, was in excess of 400 m. A net radiometer (Q7.1, Radiation Energy Balance Syst., Seattle, WA) was positioned 1.75 m above the surface. Surface temperature was measured with 0.127-mm-diam. thermocouples that had been mounted to the surface using a thin layer of epoxy. Temperature of the asphalt was measured at 0.042 m using two additional thermocouples. Soil heat flux was calculated using the combination method after assuming a linear temperature gradient between the surface and 4.2 cm. The thermal conductivity and heat capacity of the pavement were approximated as 2.1 W m^-1 K^-1 and 1.95 MJ m^-3 K^-1, respectively (Incropera and DeWitt, 1990). Signals from the gas analyzer, anemometer, and fine-wire thermocouple were sampled at 10 Hz using a Campbell Scientific CR23X data logger. All other sensors were sampled every 10 s. Raw fluxes (covariances) and scalar averages were computed and stored every 30 min using the CR23X data logger.

Postprocessing of the eddy covariance data included coordinate rotation, correction of sonic-derived H estimate (Schotanus et al., 1983), and density corrections on the raw CO₂ and water vapor fluxes (Webb et al., 1980). Theoretically, some of these corrections were not needed on the parking lot because it was almost level and latent heat flux (E) was zero. However, because the precision of the entire system was in question (measurements and processing), the postprocessing algorithms were identical to those used above vegetated surfaces. Coordinate rotation was performed using the traditional micrometeorological approach by rotating the data about two axes such that the mean vertical wind speed was zero (Paw U et al., 2000). Rotation had minimal effect on the covariances because the parking lot had minimal slope. The Schotanus et al. (1983)(Eq. [8]) humidity correction on the sonic-derived H and the Webb et al. (1980) correction on E were computed by iteration because they are interdependent. That is, the raw value of H from the sonic was used to make the initial density correction on E, and then the new E is used to make the humidity correction on H. This processing loop (Eq. [2], [3], [4], and [5]) continued until changes in H and E were negligible. After corrected values of H and E were available, the density correction was applied to CO₂ flux (Eq. [6]). Formulas used for the postprocessing sequence were

[2]

[3]

[4]

[5]

[6]

where

The last term on the right hand side of Eq. [3] was replaced with -0.072E for the calculations after assuming typical values for , , and C_p. As defined here, downward fluxes are negative, and upward fluxes are positive. Equations [2] and [6] are those derived by Webb et al. (1980). Details on the theory underlying eddy covariance are available elsewhere; a good review for the nonspecialist is provided by Baldocchi (2003).

The accuracy of the eddy covariance system and the correction procedures were evaluated by comparing the raw and corrected fluxes of E and CO₂ to known fluxes from the parking lot. Latent heat flux was assumed to be zero because the surface was dry. Surface CO₂ flux was measured in the source area of the tower using a portable gas analyzer (LI6200, LI-COR, Lincoln, NE) and hand-held chamber (Norman et al., 1992). Data were collected along a transect in the source area of the tower for 3 d during the study. The average surface CO₂ fluxes were 0.03, 0.05, and 0.02 mg m^-2 s^-1, with an average standard deviation of ±0.02 mg m^-2 s^-1. Apparently, there were very small amounts of CO₂ diffusing through the pavement at an almost constant rate. For analyses purposes, it was assumed that CO₂ flux was constant at 0.032 mg m^-2 s^-1 during the entire study.

Two additional eddy covariance systems were operated on tallgrass prairie and an 80-yr-old cedar forest. Data collected in June and July were used to examine how the density corrections on the CO₂ flux measurements would affect the calculation of the daily C balance. The eddy flux instrumentation used at these locations was identical to that used at the parking lot. Net radiation also was measured with a Q7.1 radiometer, but G was measured using heat flux plates (HFT-3, Radiation Energy Balance Syst., Seattle, WA) at 0.05 m and dual-probe heat capacity sensors at 0.025 m (Campbell et al., 1991). The prairie site was located on the Konza Prairie Natural Research Area approximately 15 km south of Manhattan, KS. The vegetation at the site was dominated by C4 grasses, including big bluestem (Andropogon gerardiiVitman) and indian grass [Sorghastrum nutans (L.) Nash]. Leaf area index of the grassland was approximately 2.5 m² m^-2 during the study. Eddy flux instrumentation was positioned 3 m above the soil surface. The forest site was an 80-yr-old stand of eastern red cedar (Juniperus virginiana L.) located 32 km north of Manhattan, KS. The average tree height was 9 m, and the eddy flux instruments were positioned at 14 m. Measurements of the within-canopy CO₂ concentration profile were used to estimate the rate change in C storage in the 0- to 14-m layer. The storage term was added to the eddy covariance estimates of F_C. At the time of the study, the region around Manhattan was experiencing an extreme drought; there had not been any measurable rainfall at either site in the 5 wk before the measurements.

For a brief period, a fine-wire thermocouple (0.013 mm diam.) was added to the prairie site and positioned near the open-path gas analyzer, approximately 15 cm from the sonic anemometer. The difference in H estimated using the sonic temperature was compared to H derived from the thermocouple to estimate the effect of sensor separation on the calculation of water vapor and CO₂ fluxes (Villalobos, 1997; Laubach and McNaughton, 1999).

In addition to the density corrections, the effect of forced energy balance closure also was evaluated using two closure techniques described by Twine et al. (2000). In the first method, it was assumed that the eddy covariance systems measured the B correctly after the water vapor and density corrections had been applied to H and E (Eq. [2]–[5]). New values for latent heat flux (E_BR) then were computed from B and Rn - G in the traditional manner. It was then assumed that errors in F_C were in the same proportion as those for water vapor. Thus, new values for CO₂ flux (F_C,BR) were obtained as the product of F_C (from Eq. [6]) and the ratio between E_BR and E. The second technique was similar to the first approach except that latent heat flux was estimated as the residual of the energy balance (E_RS = Rn - G - H). New estimates of CO₂ flux (F_C,RS) were estimated as the product of F_C (from Eq. [6]) and the ratio between E_RS and E.

	RESULTS AND DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Parking Lot Test
Data from the parking lot test were evaluated over a 5-d period from 5 July to 10 July 2002 [day of year (DOY) 186–191] and between 15 and 17 July (DOY 196–198). Approximately 15% of the data collected during these periods were discarded because of wind direction. Typical summer conditions prevailed with mostly clear skies, Rn of 15 MJ m^-2 d^-1, maximum daily air temperatures between 35 and 40°C, and wind speeds between 1 and 5 m s^-1. A large portion of the daytime Rn was conducted into the pavement (Fig. 1) . Soil heat flux, shown as a residual in Fig. 1 (G = Rn - H), was near 400 W m^-2 at midday. The values of G calculated using temperature gradient measurements were similar in magnitude. However, there was large uncertainty in these calculations because the thermal conductivity and heat capacity of the pavement was unknown. When tabled values for pavement thermal properties were used to calculate G, and E was assumed negligible, energy balance closure for the experiment was (Rn - G) = 0.72(H + E) - 6.5 W m^-2. The difficulty in measuring G from this surface decreased the utility of the energy balance closure to test. Variation in wind speeds over time was larger than observed at flux towers on nearby prairie ecosystems. Surface temperatures were between 55 and 65°C at midday, which were 20 to 25°C greater than air temperature. The surface boundary layer was unstable during the entire test, with an average daytime Obukhov stability parameter (z/L) of -0.018. Sensible heat fluxes were about 250 W m^-2 at midday, which seemed low for a surface with no latent heat flux. However, the smooth surface had a low aerodynamic conductance, and the asphalt surface was a large energy sink.

View larger version (36K): [in this window] [in a new window]   Fig. 1. Net radiation (Rn), sensible heat flux (H), and soil heat flux (G) at the parking lot used to test the open-path eddy covariance instrumentation. Soil heat flux was calculated as a residual after assuming latent heat flux was negligible, G = -(Rn - H).

View larger version (22K): [in this window] [in a new window]   Fig. 2. Comparison of sonic- and thermocouple-derived measurements of air temperature. Each point is a 30-min average.

View larger version (23K): [in this window] [in a new window]   Fig. 3. Comparison of sonic- and thermocouple-derived measurements of sensible heat fluxes as obtained by eddy covariance. Each point represents a 30-min flux measurement.

View larger version (37K): [in this window] [in a new window]   Fig. 4. Eddy covariance measurements of CO2 flux above a parking lot for 7 d in July 2002. Shown are uncorrected raw measurements and the same data after density corrections have been applied. Actual fluxes of CO2 from the surface were approximately 0.04 mg m-2 s-1 (dashed line). Downward fluxes are negative; upward fluxes are positive.

View larger version (21K): [in this window] [in a new window]   Fig. 5. Error in the 30-min measurements of CO2 flux plotted against the corresponding sensible heat flux.

View larger version (40K): [in this window] [in a new window]   Fig. 6. Eddy covariance measurements of latent heat flux above a parking lot for 7 d in July 2002. Shown are uncorrected raw measurements and the same data after density corrections have been applied. Actual latent heat fluxes were assumed to be zero. Downward fluxes are negative; upward fluxes are positive.

Carbon Dioxide Fluxes from Tallgrass Prairie and Cedar Forest
Figure 7 shows raw and corrected F_C as measured at the prairie and cedar forest sites during the same period used for the parking lot study. The region was experiencing drought, so canopy photosynthesis was less than normal and the B was comparatively large. Sensible heat fluxes at midday were 200 and 500 W m^-2 on the prairie and forest, respectively. Because H was large and F_C was small, the density correction had a dramatic effect on the diurnal measured pattern and daily total CO₂ fluxes. Midday CO₂ fluxes were reduced from -0.66 to -0.24 mg m^-2 s^-1 in the prairie, a 66% change, and from -0.92 to -0.17 mg m^-2 s^-1 in the forest, a 82% change. At the prairie site (Fig. 7a), the corrected data showed a clear midafternoon drop in canopy photosynthesis followed by a slight recovery toward sunset, a feature resulting from high light and high vapor pressure deficits (A.K. Knapp, Kansas State Univ., personal communication, 2002). In the raw data, the afternoon shutdown of photosynthesis was completely masked by the density effect. When data in Fig. 7 were integrated over 5 d, the raw and corrected daily CO₂ balances were -12.9 and -4.3 g m^-2 d^-1, respectively, for the prairie and -15.2 and 1.8 g m^-2 d^-1, respectively, for the forest. The forest is a worst case for the density effect (i.e., small F_C, large H); the raw data showed the forest was a large sink for atmospheric C, and the corrected data showed it was actually a net source.

View larger version (36K): [in this window] [in a new window]   Fig. 7. Eddy covariance measurements of CO2 fluxes from (a) tallgrass prairie and (b) cedar forest, both located near Manhattan, KS. Shown are uncorrected raw measurements and the same data after density corrections have been applied. Downward fluxes are negative; upward fluxes are positive.

Energy Balance Closure
A tendency to underestimate energy and mass fluxes has been a pervasive problem with the eddy covariance technique (e.g., Twine et al., 2000). As an example, Fig. 8 shows the degree of energy balance closure at the prairie and forest sites between DOY 174 and 182. If perfect closure had been achieved, all data would fall on a 1:1 line; however, closure was 0.79 and 0.96 for the prairie and forest locations, respectively. The forest site, which had an average daytime B of 2.4, always had better energy balance closure than the prairie site, which had a B of 0.24. This suggests that H may have been measured more accurately than E. Also, the length scale of the eddies responsible for transport was larger at the forest; thus, the frequency response and sensor separation errors may have been smaller. In both cases, there was a slight tendency for H + E to be less than Rn - G when fluxes were less than 300 W m^-2. The validity of these comparisons depends on the accuracy of the Rn and G measurements; thus, lack of closure cannot be isolated to the eddy covariance equipment alone. Errors in Rn - G are often 5 to 10%. Nevertheless, the size of the energy balance discrepancy in the prairie site was disconcerting; it suggested the fluxes of E may have been underestimated. If the covariance measurements for CO₂ were subject to the same errors as those affecting water vapor, then F_C was probably underestimated as well.

View larger version (23K): [in this window] [in a new window]   Fig. 8. Sum of sensible and latent heat fluxes (H + E) vs. available energy (Rn - G) for tallgrass prairie and a cedar forest. Data were collected between day of year 168 and 174, 2002; each point is a 30-min measurement.

View this table: [in this window] [in a new window]   Table 2. Daytime CO2 fluxes (0800–1800 h) from the tallgrass prairie between 18 June and 8 July 2002. Shown are total CO2 fluxes at different stages of a sequential correction process where FC,raw are data directly from the data logger, FC,rot are data after coordinate rotation, FC,den are data after the density correction, and FC,BR and FC,RS are data after adjustment for energy balance closure using the Bowen ratio (B)– and residual-based approaches, respectively. Also shown are the vertical rotation of the sonic anemometer (degrees), B, and the percentage of energy balance closure, (H + E)/(Rn - G) x 100.

Lack of energy balance closure, in part, could be caused by underestimates of E resulting from sensor separation and inadequate sensor response. These corrections were not applied to data in this study but are typically less than 10% and are often much smaller during the day (unstable conditions, higher wind speeds). To examine the potential errors caused by sensor separation, a 2-wk experiment was performed at the prairie site following the approach of Laubach and McNaughton (1999). Results showed that sensor separation of 0.15 m caused underestimates of flux between 0 and 3% under unstable conditions. One day of data were collected under stable conditions following an early morning rain shower. Between 0900 and 1500 h, the stability parameter z/L averaged 0.028, and sensor separation caused a 15% underestimate of flux. These results are similar to those of Laubach and McNaughton (1999). While the magnitudes of the observed sensor separation errors were significant under stable conditions, they were not large enough to account for the lack of energy balance closure shown in Table 2 (i.e., predominantly unstable conditions). Nevertheless, a more complete analysis would include a correction for these effects. For example, this correction would be more important when working with an irrigated crop when stable conditions are common.

	CONCLUSIONS

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
These experiments, like earlier studies, demonstrate how fluctuations in air density can cause a severe overestimate of downward CO₂ fluxes when using open-path eddy covariance. The Webb et al. (1980) corrections removed this effect almost completely during the parking lot experiments, and the corrected CO₂ flux measurements were within 1 g m^-2 d^-1 of independent estimates from chamber measurements. However, the noise in flux measurements, which appeared random for F_C and slightly systematic for E, was approximately ±0.1 mg m^-2 s^-1 and ±20 W m^-2, respectively (Fig. 4). This indicates that other factors, such as instrumentation errors, inadequate sampling, or failed assumptions regarding the state of the boundary layer, were affecting the measurements. Data from prairie and forest ecosystems showed that removing density effects changed estimates of CO₂ flux by 20 to 80% and in some cases changed the estimate of daily CO₂ flux at the surface from negative (sink) to positive (source).

Energy balance closure at the prairie site was about 80%, which was very similar to other reports (Aubinet et al., 2000; Twine et al., 2000). Adjustment to CO₂ flux based on forced energy balance closure greatly increased the estimates of downward CO₂ flux by 16 to 35%. However, energy balance approaches for adjusting eddy covariance data are limited because they depend on the accuracy of Rn and G, must assume that errors in CO₂ are proportional to E, must assume that advection is negligible (assume horizontal uniformity), and in the case of the B–based correction, must assume errors in H an E are proportional. Furthermore, all energy balance–based corrections in CO₂ flux will perform poorly at night when Rn - G is small.

The inherent tendency to underestimate fluxes when using eddy covariance may be linked to the errors caused by sensor separation and inadequate frequency response of the sensors. There is some debate on how to correct the data for these effects, but efforts are underway to develop a standardized approach (Massman and Lee, 2002). Unfortunately, these corrections are difficult to implement for the nonspecialist because they require calculation of cospectra using high-frequency (10 Hz) data. There is also considerable expertise and experience required to interpret the cospectra properly. Errors caused by inadequate frequency response and sensor separations are typically less than 10% but depend on wind speed, boundary-layer stability, the height of the sensors above the ground, and the type of instruments deployed. During the night, these errors are exacerbated and, along with other problems (e.g., drainage flow), greatly reduce confidence in eddy covariance measurements. All of these issues can have a profound effect on measuring the long-term C balance of a site (Goulden et al., 1996; Moncrieff et al., 1996).

Despite some problems, open-path eddy covariance is a very useful tool for examining how different agronomic practices affect field-scale processes in the soil–plant–atmosphere continuum (canopy photosynthesis, evapotranspiration, etc.). Historically, agronomic research has focused on comparing plots or fields that are under differing management regimes or treatments. Eddy covariance data from two adjacent but differently managed fields can readily identify how certain practices affect the field energy, water, and C balances. Even if the flux data were slightly biased, it is likely that the bias would be very similar at each tower location, such that the measured differences between the two fields would be quite accurate. Certainly, one can think of exceptions, but measuring differences is always easier than measuring absolutes regardless of the instruments employed. Measuring the long-term (e.g., annual) absolute C balances of fields will continue to be a challenge. However, it is probable that both advances in theory and improvements in instrumentation will progress to a point that eddy covariance can be routinely used by the nonspecialist. The concept of measuring multiple trace gas fluxes with instruments that can be deployed in a matter of minutes sounds a bit make-believe today but will most likely be a reality for the agronomist of the future.

	ACKNOWLEDGMENTS

 
Research was supported in part by grants from National Aeronautical and Space Agency, the National Institute for Global Environmental Change (Great Plains Regional Center), and the National Science Foundation. Travel support for J.L. Heilman was provided by the Big 12 Fellowship Program. F.W. Caldwell provided technical support.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Contribution no. 03-83-J from the Kansas Agric. Exp. Stn.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 

• Aubinet, M., A.Grelle, A. Ibrom, Ü. Rannik, J. Moncrieff, T. Foken, A.S. Kowalski, P.H. Martin, P. Berbigier, Ch. Bernhofer, R. Clement, J. Elbers, A. Granier, T. Grünwald, K. Morgenstern, K. Pilegaard, C. Rebmann, W. Snijders, R. Valentini, and T. Vesala. 2000. Estimates of the annual net carbon and water exchange of European forests: The EUROFLUX methodology. Advances in Ecological Research. 30:113–166.
• Baldocchi, D.D. 2003. Assessing ecosystem carbon balance: Problems and prospects of the eddy covariance technique. Global Change Biol. 9:478–492.
• 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.[ISI]
• Campbell, G.S., C. Calissendorf, and J.H. Williams. 1991. Probe for measuring soil specific heat using a heat pulse method. Soil Sci. Soc. Am. J. 55:291–293.[Abstract/Free Full Text]
• Campbell, G.S., and J.M. Norman. 1998. Introduction to Environmental Biophysics. 2nd ed. Springer-Verlag, New York.
• Desjardins, R.L., and E.R. Lemon. 1974. Limitations in the eddy covariance technique for determination of the carbon dioxide and sensible heat fluxes. Boundary-Layer Meteorol. 5:475–488.
• Goulden, M.L., J.W. Munger, S.M. Fan, B.C. Daube, and S.C. Wofsy. 1996. Measurements of carbon storage by long-term eddy correlation: Methods and critical assessment of accuracy. Global Change Biol. 2:169–182.
• Horst, T. 1999. On frequency response corrections for eddy covariance flux measurements. Boundary-Layer Meteorol. 94:517–520.
• Incropera, F.P., and D.P. DeWitt. 1990. Fundamentals of heat transfer. 3rd ed. John Wiley & Sons, New York.
• Kramm, G., R. Dlugi, and D.H. Lenschow. 1995. A re-evaluation of the Webb correction using density weighted averages. J. Hydrol. 166:283–292.
• Laubach, J., and K.G. McNaughton. 1999. A spectrum independent procedure for correcting eddy covariance eddy fluxes measured with separated sensors. Boundary-Layer Meteorol. 89:445–467.
• Leuning, R., O.T. Denmead, A.R.G. Lang, and E. Ohtaki. 1982. Effects of heat and water vapor transport on eddy covariance measurements of CO₂ fluxes. Boundary-Layer Meteorol. 23:209–222.
• Leuning, R., and M.J. Judd. 1996. The relative merits of open- and closed-path analyzers for measurement of eddy fluxes. Global Change Biol. 2:241–253.
• Massman, W.J. 2000. A simple method for estimating frequency response corrections for eddy covariance systems. Agric. For. Meteorol. 104:185–198.
• Massman, W.J., and X. Lee. 2002. Eddy covariance corrections and uncertainties in long-term studies of carbon and energy exchanges. Agric. For. Meteorol. 113:121–144.
• Moncrieff, J.B., Y. Malhi, and R.L. Leuning. 1996. The propagation of errors in long-term measurements of land–atmosphere fluxes of carbon and water. Global Change Biol. 2:231–240.
• Monteith, J.L., and M.H. Unsworth. 1990. Principles of environmental physics. Edward Arnold, New York.
• Moore, C. 1986. Frequency response corrections for eddy correlation systems. Boundary-Layer Meteorol. 37:17–35.
• Norman, J.M., R. Garcia, and S.B. Verma. 1992. Soil surface CO₂ fluxes and the carbon budget of a grassland. J. Geophys. Res. 97:18845–18853.
• Paw U, K.T., D.D. Baldocchi, T.P. Meyers, and K.B. Wilson. 2000. Correction of eddy-covariance measurements incorporating both advective effects and density fluxes. Boundary-Layer Meteorol. 97:487–511.
• Schotanus, P., F.T.M. Nieuwstadt, and H.A.R. De Bruin. 1983. Temperature measurement with a sonic anemometer and its application to heat and moisture fluxes. Boundary-Layer Meteorol. 26:81–93.
• Twine, T.E., W.P. Kustas, J.M. Norman, D.R. Cook, P.R. Houser, T.P. Meyers, J.H. Prueger, P.J. Starks, and M. Welely. 2000. Correcting eddy-covariance flux underestimates over a grassland. Agric. For. Meteorol. 102:279–300.
• Villalobos, F.J. 1997. Correction of eddy covariance water vapor flux using additional measurements of temperature. Agric. For. Meteorol. 88:77–83.
• Webb, E., G. Pearman, and R. Leuning. 1980. Correction of flux measurements for density effects due to heat and water vapour transfer. Q. J. R. Meteorol. Soc. 106:85–100.
• Wilczak, J., S. Oncley, and S.A. Stage. 2001. Sonic anemometer tilt correction algorithms. Boundary-Layer Meteorol. 99:127–150.

This Article Abstract Figures Only Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in ISI Web of Science Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via ISI Web of Science (7) Citing Articles via Google Scholar Google Scholar Articles by Ham, J. M. Articles by Heilman, J. L. Search for Related Content PubMed Articles by Ham, J. M. Articles by Heilman, J. L. Agricola Articles by Ham, J. M. Articles by Heilman, J. L. Related Collections Agroclimatology