Raymond McCord
Building 1507
PO Box 2008, MS 6407
Oak Ridge, TN 37831-6407

Telephone: (865) 574-7827
Fax: (865) 574-4665

E-Mail: mccordra@ornl.gov


This data set contains 30-minute resolution flux data from the Coordinated 
Energy and Water cycle Observation Project (CEOP) Global Energy and Water Cycle Experiment 
(GEWEX) Climate Prediction Program for the Americas (CPPA) Reference Site operated by the 
Atmospheric Radiation Measurement (ARM) Program at its Southern Great 
Plains (SGP) facility in Kansas and Oklahoma.  This data set includes 
flux observations from 23 locations.  This data set covers the time 
period 1 October 2002 through 31 December 2009.  Further information about 
the ARM SGP site is available at the 
following URL: http://www.arm.gov

1.1  Station Locations

Site          State   Elev (m)  Latitude  Longitude   Surface Type
E1_Larned      KS      632     38.202 N   99.316 W   Wheat
E2_Hillsboro   KS      450     38.306 N   97.301 W   Grass  
E3_Le_Roy      KS      338     38.201 N   95.597 W   Wheat and Soybeans (rotated)
E4_Plevna      KS      513     37.953 N   98.329 W   Rangeland (ungrazed)
E5_Halstead    KS      440     38.114 N   97.513 W   Wheat
E6_Towanda     KS      409     37.842 N   97.020 W   Alfalfa
E7_Elk_Falls   KS      283     37.383 N   96.180 W   Pasture
E8_Coldwater   KS      664     37.333 N   99.309 W   Rangeland (grazed)
E9_Ashton      KS      386     37.133 N   97.266 W   Pasture
E10_Tyro       KS      248     37.068 N   95.788 W   Alfalfa
E12_Pawhuska   OK      331     36.841 N   96.427 W   Native Prairie
E13_Lamont     OK      318     36.605 N   97.485 W   Pasture and Wheat
E14_Lamont     OK      315     36.607 N   97.488 W   Pasture and Wheat
E15_Ringwood   OK      418     36.431 N   98.284 W   Pasture
E16_Vici       OK      602     36.061 N   99.134 W   Wheat
E18_Morris     OK      217     35.687 N   95.856 W   Pasture (ungrazed)
E19_El_Reno    OK      421     35.557 N   98.017 W   Pasture (ungrazed)
E20_Meeker     OK      309     35.564 N   96.988 W   Pasture
E21_Okmulgee   OK      240     35.615 N   96.065 W   Forest
E22_Cordell    OK      465     35.354 N   98.977 W   Rangeland (grazed)
E24_Cyril      OK      409     34.883 N   98.205 W   Wheat (gypsum hill)
E26_Cement     OK      400     34.957 N   98.076 W   Pasture
E27_Earlsboro  OK      300     35.269 N   96.740 W   Pasture

Note:  E19 has a slightly different location from the start of collection through
6 May 2003 (35.549 N and 98.020 W).  From 7 May 2003 to the end of collection
the location is as above.

1.2 Time Period Covered by Data

Site           Start Date       End Date
E1_Larned       8 Mar 2004   14 Oct 2009 (data collection ended)
E2_Hillsboro    1 Oct 2002   20 Oct 2009 (data collection ended)
E3_Le_Roy      10 Mar 2004   24 Oct 2009 (data collection ended)
E4_Plevna       1 Oct 2002   31 Dec 2009
E5_Halstead     1 Oct 2003    2 Nov 2009 (data collection ended)
E6_Towanda      1 Oct 2003   31 Dec 2009
E7_Elk_Falls    1 Oct 2002   31 Dec 2009
E8_Coldwater    1 Oct 2002   10 Nov 2009 (data collection ended)
E9_Ashton       1 Oct 2002   31 Dec 2009
E10_Tyro        3 Oct 2003   31 Dec 2009
E12_Pawhuska    1 Oct 2002   31 Dec 2009
E13_Lamont      1 Oct 2002   31 Dec 2009
E14_Lamont      1 Oct 2003   31 Dec 2009
E15_Ringwood    1 Oct 2002   31 Dec 2009
E16_Vici        1 Oct 2003   31 Dec 2009
E18_Morris      1 Oct 2002   17 Nov 2009 (data collection ended)
E19_El_Reno     1 Oct 2002   31 Dec 2009
E20_Meeker      1 Oct 2002   31 Dec 2009
E21_Okmulgee   11 Feb 2004   31 Dec 2009
E22_Cordell     1 Oct 2002    1 Dec 2009 (data collection ended)
E24_Cyril      18 Mar 2004   14 Nov 2009 (data collection ended)
E26_Cement      1 Oct 2002   17 Dec 2009 (data collection ended)
E27_Earlsboro   7 May 2003    4 Dec 2009 (data collection ended)

1.3  Temporal Resolution

All data are 30-minute resolution.  See the instrumentation section for 
further information.


The ARM flux measurements are obtained from two different instrumentation
packages, the Eddy Correlation (ECOR) and Energy Balance Bowen Ratio
(EBBR) systems.

The following sites used the ECOR instrumentation:
E1, E3, E5, E6, E10, E14, E16, E21 and E24
The ECOR instrumentation provides measurements of CO2 flux, latent heat
flux and sensible heat flux.  The soil heat flux is not measured at these
stations.  The ECOR sensors are located at 3.00 m height at all stations
except E21 where the sensors are located at 21.00 m height.  These
stations started collecting data at varying times during EOP-4.

The following sites used the EBBR instrumentation:
E2, E4, E7, E8, E9, E12, E13, E15, E18, E19, E20, E22, E26 and E27
The EBBR instrumentation provides measurements of latent heat flux,
sensible heat flux and soil heat flux.  The CO2 flux is not measured at
these stations.  The soil heat flux is measured at -0.05 m height.
The sensible and latent heat flux values are shown as at a missing
(-999.99 m) height in the data file since the measurements used to
derive these parameters are at multiple heights.

2.1 ARM EBBR Instrumentation

Further information on the EBBR instrumentation is included in the ARM
EBBR Handbook:
Highlights from that document are included here.

2.1.1 Expected uncertainties

The expected uncertainty for the sensible, latent and soil heat fluxes is
10%.  They define uncertainty as the range of probable maximum deviation
of a measured value from the true value within a 95% confidence interval.

2.1.2  EBBR Components

The accuracies cited below are generally those stated by the manufacturer. 
They are sensor absolute accuracies and do not include the effects of 
system (i.e., datalogger) accuracies. Although it is not known how some 
of the manufacturers have determined sensor accuracy, it is properly the 
root square sum of any nonlinearity, hysteresis, and nonrepeatablity, 
usually referenced as percentage of full scale.

The detection limit is normally restricted to the range (sometimes called 
Calibrated Operating Range) over which the accuracy applies. In the case 
of the EBBR, some of the detection limits are those determined by the 
vendor (REBS) who performed the calibration, not by the manufacturer of 
the sensor. Some manufacturers also specify an Operating Temperature 
Range in which the sensor will physically and electronically function, 
even though the calibration may not be appropriate for use throughout that
range. When no detection limits have been listed by the manufacturer or 
the calibrating vendor, none are stated below.

Air temperature: Chromel-constantan thermocouple, Omega Engineering Inc., 
REBS Model # ATP-1, Detection Limits -30 to 40 deg C, Accuracy +/- 0.5 deg C.

Temperature/Relative Humidity Probe: Operating Temperature Range -20 to 
60 deg C. Temperature: Platinum Resistance Temperature Detector (PRTD); 
Detection Limits -30 to 40 deg C, Accuracy +/- 0.2 deg C Relative 
Humidity: Capacitive element, Vaisala Inc., Model #s HMP 35A and HMP 35D; 
Detection Limits 0% to 100% RH, Accuracy +/- 2% (0-90% RH), +/- 3% 
(90-100%), uncertainty of RH calibration +/- 1.2%.

Soil Temperature: Platinum Resistance Temperature Detector, MINCO 
Products, Inc., REBS Model # STP-1, MINCO Model # XS11PA40T260X36(D), 
Detection Limits -30 to 40 deg C, Accuracy +/- 0.5 deg C.

Soil Moisture: Soil Moisture Probe (fiberglass and stainless steel 
screen mesh sandwich), Soiltest, Inc., REBS Model # SMP-2, Soiltest 
Model # MC-300, Accuracy not specified by manufacturer (varies 
significantly depending on soil moisture and soil type).  Detection 
limits for this sensor are limited by the ability to fit a polynomial 
to the calibration data; for the SGP CART Site the detection limits 
are approximately 1% to 50% by volume.

Soil Heat Flow: Soil Heat Flow Probes, Radiation & Energy Balance 
Systems, Inc., Model #s HFT-3, HFT3.1, Accuracy not specified by 

Barometric Pressure: Barometric Pressure Sensor, Met One Instruments, 
Model #s 090C-24/30-1, Detection Limits 24 to 30 kPa; 090C-26/32-1, 
Detection Limits 26 to 32 kPa; 090D-26/32-1, Detection Limits 26 to 
32 kPa; Accuracy for all +/- 0.14 kPa.

Net Radiation: Net Radiometer, Radiation & Energy Balance Systems, 
Inc., Model Q*6.1 or Q*7.1, Accuracy +/- 5% of full-scale reading.

Wind Direction: Wind Direction Sensor, Met One Instruments, 
Model #s 5470, 020C, Detection Limits 0 to 360 deg physical (for greater 
than 0.3 ms-1 wind speed), 0 to 356 deg electrical, Accuracy +/- 3 deg.

Wind Speed: Wind Speed Sensor, Met One Instruments, model #s 010B and 
010C, Operating Temperature Range -50 to 85 deg C, Detection Limits 
0.27 to 50 ms-1, Accuracy +/- 1% of reading. Operational Limit on speed 
60 ms-1.

Datalogger: Campbell Scientific, Inc., Model CR10, Detection Limits vary 
by voltage range selected, Accuracy +/- 0.1% of full scale reading. 

2.1.3  EBBR System Configuration

The meteorological observations made with the EBBR system are:

    Air temperature at two heights (1 m separation)
    Relative humidity at two heights (1 m separation)
    Net radiation (at 2 m typical)
    Soil moisture at 2.5 cm depth
    Soil heat flow at 5 cm depth
    Soil temperature, integrated 0 to 5 cm
    Barometric pressure
    Wind direction at 2.5 m
    Wind speed at 2.5 m
    Reference temperature of control box 

The EBBR sensors (except for soil probes) are mounted on a triangular 
pipe framework that sits on the soil surface. The net radiometer mount 
extends from the south end of the EBBR frame.

A unique aspect of the system is the automatic exchange mechanism (AEM), 
which helps to reduce errors from instrument offset drift. The AEM extends 
from the north end of the frame. Aspirated radiation shields (which house 
the air temperature and relative humidity probes) are attached to the AEM. 
The openings of the aspirated radiation shields face north to reduce 
radiation error from direct sunlight.

The soil probes are buried just outside the view of and in an arc to the 
south of the net radiometer.

Heights of individual sensors are listed in Primary Quantities Measured 
with System. Heights of air temperature and relative humidity sensors vary 
from site to site and are dependent on vegetation height; these heights 
can vary seasonally as vegetation height changes.

The reference temperature sensor, barometric pressure sensor, datalogger, 
storage module, and communication equipment are located in the control 
box, which is attached at the northeast corner of the EBBR frame.

The local area of influence upon Bowen ratio measurements is contained 
within a horizontal distance of approximately 20 times the height of the 
top aspirated radiation shield on the AEM. This distance varies among 
the different extended facilities and for different times of the year 
because of differences in vegetation height, and therefore the height 
at which the AEM is installed.

The manufacturer's (REBS) name for these systems is SEBS (Surface Energy 
Balance System) ; this is the name that appears in their systems 

2.2 ECOR Instrumentation

Detailed information on the ECOR instrumentation is included in the 
ECOR Handbook
developed by ARM.  Selected highlights from that document are included here.

2.2.1  Uncertainties

The expected uncertainties are as follows:

Sensible heat flux - 6%
Latent heat flux - 5%
CO2 flux - 4%

2.2.2 ECOR Components

Ultrasonic anemometer: WindMaster Pro by Gill 
Instruments Ltd

Open-path CO2/H2O IRGA: LI-7500 by LI-COR, Inc.

2.2.3 ECOR System Configuration

In a typical arrangement, the ECOR system is placed on the north side of a wheat field; sonic and IRGA
sensor heads are mounted on a small tower at 3 m above ground level, at the end of a horizontal boom
pointing south. The computer and communication devices are installed in an enclosure with basic
temperature control (ventilation or heating). One exception to the usual arrangement is the Okmulgee site
(E21), where the ECOR system is installed on a tall tower (15 m above ground, about 3 m above the
canopy) in a hardwood forest.

The IRGA provides fast-response measurements of water vapor density and CO2 concentration in digital
and analog form; the sonic anemometer provides three wind components and the SOS data in digital form
(retrieved via serial link) at a rate of 10 Hz and performs synchronous digitization of the IRGA analog
outputs. The digital data stream from the IRGA is also recorded by the data acquisition computer; it is
used to extract IRGA diagnostics values and as a second copy of the water vapor density and CO2
concentration data.

The raw data stream is recorded into raw data files by 30-min portions and is processed every half hour
by the ECOR computer. The raw and processed data files are transferred to the Central Facility for ingest
(conversion into the netCDF format and incorporation of QC flags) and shipment to the ARM Archive.


3.1  ARM Data Collection and Processing

Full information on the ARM data collection and processing is 
available in the 
EBBR Handbook
and the
ECOR Handbook.

3.2  UCAR/JOSS Data Processing

The University Corporation for Atmospheric Research/Joint Offfice for
Science Support (UCAR/JOSS) converted the data from the raw format
provided by ARM into the CEOP EOP-3 data format agreed to by
the CEOP Scientific Steering Committee.  This format is described
in detail as part of the CEOP Reference Site Data Set Procedures
Report which is available at the following URL:


As part of this conversion, JOSS multiplied the flux values from both the
EBBR and ECOR systems by -1.0 so the sign conventions match the remainder 
of the CEOP flux data sets.  Otherwise JOSS takes the 30 minute observations 
from the EBBR and ECOR systems without change.


4.1  ARM Quality Control Procedures

For detailed information on EBBR quality control procedures please
see the
EBBR Handbook
and the
ECOR Handbook.

4.2  UCAR/JOSS Quality Control Procedures

UCAR/JOSS converted the ARM QC flags into the CEOP QC flags in the 
following manner.  If a parameter failed one of the ARM QC checks it was 
flagged as Questionable/Dubious ("D") and if it failed two or more ARM 
QC checks it was flagged as Bad ("B").  Additionally, ARM issues Data
Quality Reports (DQRs) anytime problems are noticed within the data stream
(e.g. failing instruments, calibration periods, etc).  UCAR/JOSS has 
examined the DQRs issued by ARM over this time period (over 140 DQRs were
issued) and determined when the parameters included within this data set
may have been impacted and flagged the data either "D" or "B" based on the
description of the problem included in the DQR.  The ARM DQR's are provided
as part of the data set:

Additionally, UCAR/JOSS conducted two primary quality assurrance/control
procedures on the reference site data.  First the data has been 
evaluated by a detailed QA algorithm that verifies the format is 
correct, examines any QC flags, and conducts basic checks on data 
values.  Second, JOSS conducts a manual inspection of time series 
plots of each parameter.  Additional data quality flags are applied
during this stage.


No gap filling procedures were applied to these data by either


6.1  EBBR Typical Problems to Watch For

The EBBR system is capable of producing estimates of 30-minute average 
sensible and latent heat fluxes accurate to approximately +/- 10% of the 
estimated value or +/- 10 watts per meter squared, whichever is larger, 
at the 95% confidence level. Offset and calibration drift errors in the 
measurements of the temperature and relative humidity gradients are 
significantly reduced by the use of the AEM, which switches the 
positions of the upper and lower sensors every 15 minutes.

There are a number of conditions under which the primary measurements 
may be incorrect. The data user should examine the data quality flags 
(described in Automated Quality Control (flagging contained within 
netCDF files)) and data quality reports (DQR) to determine whether 
significant malfunctions have occurred in the EBBR system. The more 
frequent sources of error are described briefly here.

Generally, when the AEM is not functioning properly, the sensible and 
latent heat flux estimates are unreliable and should not be used.  
Examine the data quality flags for AEM home signals to ensure that only 
flux data produced when the AEM is functioning properly is used.

Another source of error in the sensible and latent heat flux estimates 
lies in net radiation measurement errors. An error in the net 
radiation results in the same percentage error in the heat flux estimates.

Condensation or frost sometimes forms on the upper net radiometer dome 
(called a "windshield" by the manufacturer) at night, especially when the 
relative humidity is high, wind speeds are very low, and the sky is 
clear (allowing rapid radiational cooling). This can persist until early 
afternoon, especially on the side of the net radiometer that is away 
from the sun. If the net radiometer dessicant is not in good condition, 
condensation can also occur inside the dome during the nighttime and can
persist until early afternoon. Dessicant in poor condition can desorb 
water vapor into the dome area during the nighttime portion of the 
natural diurnal "breathing" process through the dessicant; some air 
passes through the dessicant and into the domes as the ambient 
temperature decreases with the approach of nightfall. Conversely, in the 
morning, the air in the domes warms and some moves through the dessicant 
in the opposite direction, thereby reducing the water vapor content of 
the air. This "breathing" process is intended to keep the air inside the 
domes dry. Condensation or frost inside or outside the domes can cause 
a decrease in net radiation to be measured during the day and an increase 
at night. During the day, the condensation or frost obstructs the passage 
of solar and infrared radiation through the top dome, thereby causing too 
low a value to be measured. At night, the condensation or frost prevents 
the cooling of the upper sensing surface of the net radiometer, thereby 
resulting in too large a net radiation value. The nighttime error in net 
radiation has, in extreme conditions, been measured to be as much as 
30 watts per meter squared. Within a year of installation of the first 
EBBR systems it was obvious that the dessicant was degrading and had to
be replaced more frequently than should have occurred. The cause was 
found to be a poor seal of the domes and the sensor head, a result of a 
poor sealing design. During the nighttime, undessicated ambient air would 
leak into the dome area around the seals.  The manufacturer subsequently 
modified the design, resulting in fewer occurrences. Furthermore, a 
procedure to test the seal on the domes every 3 months and at each dome 
change was instituted.

The polyethylene domes (particularly the upper one) become cloudy from 
exposure to sunlight, most rapidly in the first several months of use. 
After 6 months, the vendor states that degradation may result in as much 
as a -5% error in net radiation during the day, but with a negligible 
error at night. The domes are scheduled to be replaced every 3 months 
to minimize this error.

At first, the spare domes obtained from the manufacturer appeared to vary 
significantly in thickness, a possible small source of error (within the 
accuracy stated by the manufacturer). The nature of the manufacturing 
process (forming polyethylene hemispheres from a sheet of material) lends 
itself to some variation in thickness of the final product. The 
manufacturer was subsequently asked by ARM to inspect future domes for 
abnormalities before they are sent to the SGP CART Site.

The net radiation measurement is also affected by the amount of cooling 
produced by wind speeds of various magnitudes. The cooling effect is 
greatest on the top portion of the net radiometer (whose surface is 
warmest during the day because of solar radiational heating), primarily 
during daylight hours. After April 1996, a correction to net radiation, 
based on wind speed, was included in the CR10 software used in each EBBR 
calibration changeout kit. The changeouts were completed in November 
1996. In 1997, specially designed ventilators are to be added to the net 
radiometers. The ventilators provide a non-wind-dependent airflow, the 
effect of which is easily included in the CR10 programming. This constant 
airflow will reduce the number of occurrences of condensation and frost 
on the net radiometer domes, thereby reducing the amount of incorrect 
net radiation data at night and in the morning.

Furthermore, an error occurs when the Bowen ratio has a value near -1. 

Another source of error can be possible inaccuracies of the soil 
measurements (soil temperature change with time, soil moisture, soil heat 
flow at 5 cm) used to calculate estimated soil heat flux; aging of the 
sensors with time can result in changes in calibration or offset. When 
such changes become significant, the standard procedure is to replace 
the sensor and submit a DQR indicating the amount of error involved. 
Rarely is the error sufficiently large to declare the sensible and latent 
heat fluxes incorrect. The soil heat flux is normally a small fraction of 
the total energy exchange at the surface. Inaccuracies in the soil heat 
flux estimates normally cause less than a +/- 5% error in the sensible 
and latent heat fluxes. However, a total failure of a soil sensor (this 
is always reported in a DQR) often requires that the sensible and latent 
heat fluxes be recalculated using soil heat flux determined from the 
functioning sets of soil sensors only. The information needed to perform 
the recalculation is provided in the DQR. The alternative is to not use 
the sensible and latent heat flux estimates when a DQR has reported a soil
sensor failure.

One condition that may be hard for the data user to detect involves the 
failure of one of the air temperature thermocouples from which temperature 
gradient is determined. A failed thermocouple will typically result in 
offscale top and bottom air temperatures, thereby producing a zero 
temperature gradient and no sensible heat flux. The latent heat flux 
that is reported is then incorrect. It is possible for the sensible and 
latent heat fluxes to be recalculated using the relative humidity probe
temperatures in place of the thermocouple temperatures; this method may 
add a small amount of error to the resulting fluxes, but should produce 
fluxes that are within 20% of those that would have been calculated 
using the thermocouples.

Another condition that needs to be "looked out for" is rapid changes in 
net radiation. Rapid changes during the basic measurement averaging time 
of 12 minutes per quarter hour can result in non-linear changes that the 
30 second sampling interval does not handle well; in other words, the 
average of one or all of net radiation, temperature gradient, and vapor 
pressure gradient do not reflect a correct average of the rapid changes; 
this is a result of undersampling. Normally, this condition is difficult 
to distinguish from normal variations, but on occasion it has been 
pronounced, even causing an unexpected sign of one heat flux (usually 
sensible) and a very large value for the other (usually latent). This 
condition most commonly occurs with saturated soil and partly cloudy 


To support the continuation of this program, please include the
following 'credit line' in the acknowledgments of your

"Data were obtained from the Atmospheric Radiation Measurement
(ARM) Program sponsored by the U.S. Department of Energy, Office
of Science, Office of Biological and Environmental Research,
Environmental Sciences Division."


Further information on ARM SGP EBBR and ECOR instrumentation (including QC, 
calibration, maintenance, theory of operation, and references) can be 
found in the
EBBR Handbook
and the
ECOR Handbook.