The turbulent fluxes of heat and water vapor in the atmosphere above the canopy were obtained during FIFE using a network of ground-based stations employing either eddy correlation or Bowen ratio measurement techniques. These stations were distributed across the FIFE landscape according to a topographic and land-use stratification. In 1987, 22 stations collected data at 20 sites. At two of these sites, a Bowen ratio and an eddy correlation station were located together. In 1989, 16 stations were operated at 13 sites. Two of the 13 were collocated Bowen ratio and eddy correlation sites. In addition, a portable Bowen ratio system and moveable eddy correlation gear were moved between sites in 1987 to provide comparison data.

The surface flux data set is comprised of data obtained by eddy correlation and Bowen ratio stations, as well as a standardized data set of surface fluxes that are compared to model projections. Two additional data sets are found in the "Grab Bag" section of this CD-ROM. One contains the Flux data and the Radiation data averaged each half-hour over selected stations for the four Intensive Field Campaigns (IFCs). The other is from the moveable eddy correlation gear, which was deployed at multiple sites during 1987.

Five investigators obtained the eddy correlation data:

      ID       Investigator

ECA     M. Wesley (Argonne National Laboratory)

ECV     S. Verma  (Univ. of Nebraska, Lincoln)

ECW     H. Weaver (U.S. Geological Survey)

ECB     J. Stewart (Institute of Hydrology, U.K.)

ECG     R.T. Field (Univ. of Delaware)

Surface flux measuring stations equipped with eddy correlation systems observed the turbulent exchange of sensible and latent heat above the canopy by measuring the co-variance of the vertical wind velocity with respectively the air temperature and either the water vapor density or water vapor pressure. Some stations also observed the flux of momentum, carbon dioxide, and ozone. Since turbulent fluxes are assumed to be nearly constant within the atmospheric boundary layer within 10 to 100 m of the surface, these observations are assumed to represent surface values. The flux observations are one-half-hour averages for all stations except stations 26 and 926 (8739-ECB) which reported one-hour values. These stations also usually observed the net radiation and ground heat flux, and reported statistics of the vertical wind velocity, temperature, and humidity fluctuations. Some stations also observed solar radiation, mean wind speed and direction, precipitation, and surface brightness temperature.

The four terms of half-hour values of the energy budget are found in the entries SOIL_HEAT_FLUX (ground-heat flux at the soil surface), R_NET (net radiation), LATENT_HEAT_FLUX and SENSIBLE_HEAT_FLUX (eddy correlation determination at the height of the eddy correlation sensors). Observations in SOIL_TEMP_1, SOIL_TEMP_2, etc. give half-hour averages of soil temperature. These may be at different depths or may be separately observed temperatures for the same depth (see investigator's documentation). Most investigators used soil temperature at 2.5-cm depth in conjunction with soil bulk density and moisture content as reported in the Gravimetric Soil Moisture data set (see the Soil Moisture Summary Document) to compute the change in heat stored per half-hour in the top 0-5 cm layer of the soil. This value is presented in HEAT_STORAGE. Most investigators also observed the soil heat flux directly at 5-cm depth which is reported in SOIL_HEAT_FLUX_1. Some investigators reported separate heat flux observations at the same depth (SOIL_HEAT_FLUX_2, SOIL_HEAT_FLUX_3) which should be averaged for a more representative estimate while others reported a value previously averaged from several observations.

To supplement the eddy correlation observations, measurements of wind speed, temperature, vapor pressure, and solar radiation were also made by some investigators. Check the detailed data set documentation to determine what each investigator reports and how the measurements may be used. Additional information about observations at individual sites will be found in Kanemasu, et al. (1992) and the surface flux index file provided with the documentation files on this CD-ROM.

The Bowen ratio observations were obtained by six investigators:

      ID       Investigator (Original Institution)

BRS     E. Smith  (Florida State University)

BRK     G. Asrar  (Kansas State University)

BRV     S. Verma  (Univ. of Nebraska, Lincoln)

BRL     L. Fritschen (Univ. of Washington)

BRW     H. Weaver (U.S. Geological Survey)

BRG     B. Kustas (U.S.D.A. Agricultural Research Service)

Surface flux measuring stations equipped with Bowen ratio systems inferred the fluxes of sensible and latent heat above the canopy by measuring and partitioning the available energy. The available energy is the algebraic sum of the net radiation and the soil heat flux. It is partitioned between sensible and latent heat flux according to the ratio of careful measurements of the difference in air temperature between two heights to the difference in water vapor pressure between the same two heights. Since turbulent fluxes are assumed to be nearly constant within the atmospheric boundary layer within 10 to 100 m of the surface, these observations are assumed to represent surface values.

Of necessity, these surface flux stations observed net radiation and soil heat flux, and vapor pressure and air temperature at two levels. Some stations also observed wind speed and direction, solar radiation, long-wave atmospheric radiation, surface brightness temperature, and precipitation.

The four terms of half-hour values of the energy budget are found in SOIL_HEAT_FLUX (ground-heat flux at the soil surface), R_NET (net radiation), LATENT_HEAT_FLUX and SENSIBLE_HEAT_FLUX (Bowen ratio determination at the height of the sensors). Observations in SOIL_TEMP_1, SOIL_TEMP_2, etc. give half-hour averages of soil temperature. These may be at different depths or may be separately observed temperatures for the same depth (see investigator's documentation). Most investigators used soil temperature at 2.5-cm depth in conjunction with soil bulk density and moisture content as reported in the Gravimetric soil moisture data set (see the Soil Moisture Summary Document) to compute the change in heat stored per half-hour in the top 0-5 cm layer of the soil. This value is presented in the table HEAT_STORAGE. Most investigators also observed the soil heat flux directly at 5-cm depth which is reported in the table SOIL_HEAT_FLUX_1. Some investigators reported separate heat flux observations at the same depth (SOIL_HEAT_FLUX_2, SOIL_HEAT_FLUX_3) which should be averaged for a more representative estimate while others reported a value previously averaged from several observations.

Additional information about observations at individual sites will be found in Kanemasu, et al. (1992) and the surface flux index file provided with the documentation files on this CD-ROM.

A standardized surface heat flux data set produced from the net radiation, latent heat, sensible heat, and ground heat flux data acquired at each flux station during FIFE in 1987. Since erroneous jumps routinely occur in the sensible and latent heat flux time series during the early morning and evening hours; a series of tests was developed to flag the times of these jumps as well as times when energy imbalances occur. An additional flag is included based on model results using the Simple Biosphere (SiB) model. This flag provides an indication of the reasonableness of the fluxes for a station during each intensive field campaign (IFC) in 1987.

Data from 17 surface flux stations were used to generate averages for four variables: RNet, LH, SH, and Soil heat flux. Six of these stations were eddy correlation sites and the remaining 11 were Bowen ratio stations. The short-wave and long-wave radiation data includes all available data for SolDn, SolRef, LWDN, and LWUP, including data from the 17 stations used in generating the surface flux averages and five additonal stations.

Ten locations (at which stationary systems were installed for many months) were instrumented in 1987 for a few days at a time with a portable Bowen ratio system. This roving system visited all but the southeast quadrant of the FIFE study area. Similarly, a moveable eddy correlation setup was deployed at eight of the stationary system sites during the 1987 IFCs. On 11-12 July, 1987, both roving systems were co-located 25 m SSW of the 10-m tower (2731-ECA). The eddy correlation roving station data is not archived in FIFE standard format with the rest of the surface flux data, but is located in the "Grab Bag" section of this CD-ROM, as received from the investigator.

A brief introduction to radiation instruments is offered to clarify the various radiation observations reported. Fritschen and Gay (1979) provide further detail.

• Net Radiometer: Measures the difference between the incoming and outgoing hemispheric radiation on a horizontal plane over the 0.4 to 40 micrometer waveband.
• Pyranometer: Measures the intensity of the direct solar plus the diffuse radiation from the sky hemisphere on a horizontal surface. The glass filter of its hemispheric wind screen limits the instrument sensitivity to the waveband 0.4 to 4.0 micrometers.
• Pyrgeometer: Measures the intensity of thermal (or long-wave) radiation from the sky hemisphere. The sensor output must be corrected for the effect of the temperature difference between of the hemispheric infrared filter (which excludes incident radiation with wavelength less than 3.6 micrometers) and the instrument reference thermal sink.
• Pyrheliometer: Measures the intensity of the direct beam of solar radiation. A thermal sensor views the solar disk through a tube. If the instrument is not equipped with a filter or windscreen it is sensitive to the waveband roughly 0.35 to 40 micrometers.

Refer to the recent reviews by Dabberdt, et al. (1993) and Verma (1990) for general background on the techniques for measuring atmospheric fluxes employed here.

The observed eddy fluxes apply to an upwind surface "footprint" which is spread over a range between perhaps 20 to over 100 times the height of the instrument sensors (see Leclerc and Thurtell (1989), and Schmid and Oke (1990).

For the flux observations to be validly related to the upwind surface conditions the fetch must be unobstructed by obstacles to the wind flow, must be over level terrain so that the mean horizontal wind is not subjected to vertical accelerations, and the surface should be uniform over the footprint.

The sensors must be sufficiently high that the significant flux carrying wind fluctuations are large enough to be properly sensed by the sensors, yet not so high that the fetch requirements cannot be satisfied. Each investigator exposed his eddy correlation sensors according to conditions at his site. A list of eddy correlation stations and 1987 sensor heights above the ground follows:

      Grid ID          Site Nos.       1987 Height        1989 Height


2133-ECA           906                                 2.55 m

2731-ECA             4             12.3 m

4439-ECV         16, 916          2.25-2.5 m

4609-ECW            22            1.2-2.03 m

8739-ECB         26, 926            2.5 m

6943-ECW            28            0.9-1.15 m

4268-ECG            30              1.85 m

These data constitute the primary observations of surface fluxes for which the FIFE seeks to develop techniques for remote estimation. The eddy correlation and Bowen ratio data should be considered as complementary sets of observations of the same fluxes. These two approaches are the most widely used techniques for surface flux measurements. These data provide an important opportunity to compare the two approaches in an operational setting. Neither approach can be demonstrated as unequivocably superior to the other in the FIFE setting.

Members of the surface flux group worked together to minimize spurious instrumental and experimental influences on the observations. Kanemasu, et al. (1992) describe these efforts. All Bowen ratio investigators employed the same approach to obtain the ground heat flux. All eddy correlation investigators who observed ground heat flux except Station 16 also used this approach. Common methods for computing all parameters, such as air density and heat capacity and soil volumetric heat capacity were adopted. Radiometer calibration, comparison, and adjustments are described by Field, et al. (1992). Comparison of complete surface flux stations by means of roving installations and co-located stations is described by Nie, et al. (1992b) and Fritschen, et al. (1992).

These efforts establish a conservative basis for specifying level of agreement which could be expected among the observations in the absence of site and exposure differences.

A number of other FIFE data sets contain closely related data which augment the observations discussed here. Automated micrometerorological stations monitored (from May, 1987 through October, 1989, at half-hourly intervals) wind speed, solar radiation, reflected solar radiation, net radiation, PAR, incoming long-wave radiation, surface brightness temperature, wet bulb and dry bulb air temperature, humidity, rainfall, soil temperature, and soil infrared temperatures which can be used in conjunction with similar observations from surface flux sites. The data sets collected by the University of Nebraska surface radiation and biology investigators also contain solar radiation, surface infrared temperature, air temperature, and humidity at certain inoperative AMS stations during portions of 1987. Long-wave sky radiation for these situations is also reported.

Information on soil properties, such as bulk density and soil moisture, is described in the soil moisture data sets summary document on this CD-ROM. The vegetation data sets (see the biological characterization summary document) contain information on canopy height, soil depth, live and dead biomass, green leaf area index, and rates of plant growth which are needed to intepret individual site radiation and heat flux observations. Radiances observed at the surface flux sites from various platforms are presented are described in the surface reflectance summary document, the satellite data documents, and the summary document describing site specific reflectances derived from satellite and aircraft imagery.

Kanemasu, et al. (1992) believe that the surface flux observations constitute a robust and reliable set of measurements. However, it should be noted that care must be shown in the use of individual flux observations. The following considerations should be kept in mind when using these data.

Eddy correlation observations taken under ideal site conditions are typically subject to 10% to 20% variability between consecutive half-hour observations largely because of variability in the turbulence structure. Eddy correlation stations which also observe net radiation and ground heat flux provide all four terms of the energy budget. They would be expected to sum to zero so that examining the energy budget residuals could provide a test of overall measurement accuracy.

However, these residuals for half-hour averages frequently depart from zero by 10% to 20% of the net radiation, although the residuals averaged over all daylight hours may be considerably smaller. While some of this variability may arise from error in either the net radiation or ground heat flux, the largest contributors to the variability are the eddy flux observations themselves. Some of the observed eddy flux variability may arise from advection due to un-level or non-uniform surface conditions along the fetch. Most of the variability is probably inherent in the turbulence structure itself. Consequently, it is difficult to reduce the claimed uncertainty in turbulence flux observations to less than approximately 10% to 20% of the observed values.

The FIFE provided a significant opportunity for long-term comparison of different systems for flux measurement. Smith, et al. (1992b) examined the whole surface flux data set for consistency and found during IFCs 1, 2, and 3 no evidence of pronounced station bias or outliers. The two locations with co-located Bowen ratio and eddy correlation observations showed no evidence of bias between the two techniques. During IFC 4, however, they show Bowen ratio values of sensible heat flux to be 29% and 19% higher and of latent heat flux to be slightly smaller than eddy correlation values observed at the two sites. Papers by Fritschen, et al. (1992), Nie, et al. (1992b), and Moncrief, et al. (1992) report on the several systematic comparisons made among eddy correlation systems and against Bowen ratio flux systems by the FIFE surface flux group.

All of the Bowen ratio stations employed some mechanism to standardize the vapor pressure observations made by the upper and lower level. The apparatus used to obtain the observations reported by the BRS and BRG groups drew air alternately from upper and lower level to a single dew point sensor on a roughly six-minute cycle. Care had to be taken to avoid the effects of condensation in the sampling tubes. The other stations used mechanical arrangements to actually exchange the upper and lower sensors for wet and dry bulb temperatures on roughly a six-minute cycle.

It is extremely difficult to evaluate the overall accuracy of Bowen ratio flux observations because the technique allows no independent observation of the four energy balance terms. Any error in the measurement of the available energy or of either the temperature difference or vapor pressure difference (due to upwind surface inhomogeneities affecting the wind profile) will be silently folded into both calculated heat fluxes. This advection error will likely vary with wind direction. Although there is considerable variability inherent in the turbulence structure, in the absence of clouds there is less variability in the mean quantities of temperature difference and vapor pressure difference used in calculating the Bowen ratio. Consequently, Bowen ratio observations during daylight hours under cloud-free skies frequently show less half-hour-to-half-hour variability than do eddy correlation observations. However, the temperature gradient near sunset and sunrise is prone to instabilities which may lead to erroneous or impossible values of calculated fluxes. Caution is required also when using nighttime observations during times of strong temperature inversions.

The variability of surface cover and soil conditions make the soil heat flux highly variable over small distances. As an example of a short canopy situation, simultaneous values of soil heat flux at 5-cm depth observed under clear skies on one day during IFC 2 at three locations within a one-meter-square area at site 32 showed standard deviations of 20% to 40% of the values. The resulting sampling variability combined with uncertainties in sensor depth and response to the soil thermal field, and the uncertainty in estimating the change in surface layer heat storage, suggests the reported soil heat flux observations have an uncertainty on the order of 30% of the observed value.

Bowen ratio flux determinations cannot be more accurate than the observations of available energy. In spite of considerable effort to remove the biases observed among net radiometers as discussed in Field, et al. (1992) the uncertainty in the daytime net radiation observations remains between 5% and 10%. At night the uncertainty is larger. Thus, an estimate of the uncertainty in available energy when the ground heat flux is 10% to 20% of net radiation is about 10% to 25% of the available energy, increasing as the significance of the ground heat flux increases. Nie, et al. (1992b) find differences between instruments of 10%, 20%, and 30% for daily average values of respectively the net radiation, latent heat flux and the Bowen ratio. Fritschen, et al. (1992) illustrate the magnitude of the differences in estimated available energy that can be obtained at a common site. After including the uncertainty in the Bowen ratio determination and errors due to non-homogeneous site conditions, we estimate the overall uncertainty in individual flux determinations by the Bowen ratio method to be between 10% to 30% of the fluxes. The uncertainty increases as the relative contribution of the ground heat flux increase.

The two data flags included in this data set provide the user with information about errors in the data as well as the reasonableness of the fluxes based on the results of the Simple Biosphere model (SiB). Error flags were assigned to each station and observation time to spare the user the tedious task of checking the surface flux data for obvious errors and energy imbalances. The problems that SiB encountered modeling these fluxes are highlighted by the model flag. This flag provides information about the model's ability to reproduce the observed sensible and latent heat flux and where the model results suggested a bias in the station data or deviated from current theory. This information is valuable for comparison purposes with future modeling attempts.

Several strong conclusions have been drawn from these observations. Fritschen and Qian (1992) show the average evapotranspiration during the IFCs, excepting October 1987, to be about 4 mm/day. Reports by Stewart and Verma (1992) and Smith, et al. (1992b) show that latent heat flux during the growing season in 1987 was not significantly related to the green leaf area index, which ranged over more than a factor of two. The reason for this is not clear. There was a surprising degree of agreement in net radiation, sensible and latent heat fluxes across the FIFE domain. The standard deviation of one-half-hourly values of all sites on clear days was about 10% of the value for net radiation, 20% for latent heat flux, and 40% for sensible heat flux during IFCs 1, 2, and 3. Smith, et al. (1992) show that cloudiness makes the largest contribution to the flux variance among sites at a given time. Differences in energy budgets between sites in the absence of clouds from differences in topography are discussed by Nie, et al. (1992a) and Smith, et al. (1992b). Differences due to vegetation state and management are discussed by Smith, et al. (1992), Gao, et al. (1992), and Verma, et al. (1992). The evaporative fraction, the ratio of latent heat flux to the sum of latent and sensible heat fluxes (or the available energy) was found during 1987 by Shuttleworth, et al. (1989), Verma, et al. (1992), and Crosson and Smith (1992) to be highly conservative at FIFE sites during the approximately four hours before and after local noon. This finding suggests that a single daily value of the energy partitioning can be used with hourly values of available energy to obtain sensible and latent heat flux values throughout most of the daylight hours.

Crosson W.L., and E.A. Smith. 1992. Estimation of surface heat and moisture fluxes over a prairie grassland 2. J. Geophys. Res. 97(D17):18583-18598.

Dabberdt, W.F., D.H. Lenschow, T.W. Horst, P.R. Zimmerman, S.P. Oncley, and A.C. Delany. 1993. Atmospheric-surface exchange measurements. Science 260:1472-1481.

Field, R.T., L.J. Fritschen, E.T. Kanemasu, E.A. Smith, J.B. Stewart, S.B. Verma, and W.P. Kustas. 1992. Calibration, comparison, and correction of net radiometer instruments used during FIFE. J. Geophys. Res. 97(D17):18681-18695.

Fritschen, L.J., and L.W. Gay. 1979. Environmental Instrumentation. Springer-Verlag. pp. 216.

Fritschen, L., and Qian, P. 1992. Variation in energy balance components from six sites in a native prairie for three years. J. Geophys. Res. 97(D17):18651-18661.

Fritschen, L.J., P. Qian, E.T. Kanemasu, D. Nie, E.A. Smith, J.B. Stewart, S.B. Verma, and M.L. Wesley. 1992. Comparisons of surface flux measurement systems used in FIFE 1989. J. Geophys. Res. 97(D17):18697-18713.

Gao, W., M.L. Wesley, D.R. Cook, and R.L. Hart. 1992. Air-surface exchange of H2O, CO2, and O3, at a tallgrass prairie in relation to remotely sensed vegetation indices. J. Geophys. Res. 97(D17):18663-18671.

Kanemasu, E.T., S.B. Verma, E.A. Smith, L.J. Fritschen, M. Wesley, R.T. Field, W.P. Kustas, H. Weaver, J.B. Stewart, R. Gurney, G. Panin, and J.B. Moncrieff. 1992. Surface flux measurements in FIFE: an overview. J. Geophys. Res. 97(D17):18547-18555.

Leclerc, M.Y., and G.W. Thurtell. 1989. Footprint prediction of scalar fluxes using a Markovian analysis. Boundary-Layer Meteorology 52:247-258.

Moncrief, J.B, S.B. Verma, and D.R. Cook. 1992. Intercomparison of eddy correlation carbon dioxide sensors during FIFE 1989. J. Geophys. Res. 97(D17):18725-18730.

Nie, D., E.T. Kanemasu, L.J. Fritschen, L.H. Weaver, E.A. Smith, S.B. Verma, R.T. Field, W.P. Kustas, and J.B. Stewart. 1992. An intercomparison of surface energy flux measurement systems used during FIFE 1987. J. Geophys. Res. 97(D17):18715- 18724.

Nie, D., T. Demetriades-Shah, and E.T. Kanemasu. 1992. Surface energy fluxes on four slope sites during FIFE 1988. J. Geophys. Res. 97(D17):18641-18649.

Schmid, H.P., and T.R. Oke. 1990. A model to estimate the source area contributing to surface layer turbulence at a point over patchy terrain. Quart. Jour. Roy. Met. Soc. 116:965-988.

Sellers, P.J., M.D. Heiser, and F.G. Hall. 1992. Relations between surface conductance and spectral vegetation indices at intermediate-length (100 m*m - 15 km*km) scales. J. Geophys. Res. 97:19033-19059.

Smith, E.A., W.L. Crosson, B.D. Tanner. 1992a. Estimation of surface heat and moisture fluxes over a prairie grassland 1. J. Geophys. Res. 97(D17):18557-18582.

Smith, E.A., A.Y. Hsu, W.L. Crosson, R.T. Field, L.J. Fritschen, R.J. Gurney, E.T. Kanemasu, W.P. Kustas, D. Nie, W.J. Shuttleworth, J.B. Stewart. S.B. Verma, H.L. Weaver, and M.L. Wesley. 1992b. Area-averaged surface fluxes and their time-space variability over the FIFE experimental domain. J. Geophys. Res. 97(D17):18599-18622.

Shuttleworth W.J., R.J. Gurney, A.Y. Hsu, and J.P. Ormsby. 1989. FIFE: The variation on energy partition at surface flux sites. IAHS Publ. 186, pp. 67-74. Int. Assoc. of Hydrol. Sci., Wallingford, Oxfordshire, England.

Stewart, J.B, and S.B. Verma. 1992. Comparison of surface fluxes and conductances at two contrasting sites within the FIFE area. J. Geophys. Res. 97(D17):18623-18628.

Verma, S.B. 1990. Micrometeorological methods for measuring surface fluxes of mass and energy. Remote Sensing Reviews 5:99-115.