¶ Eddy Correlation Systems (ECOR)
¶ Background
The eddy correlation (ECOR) flux measurement system provides in situ, half-hour measurements of the surface turbulent fluxes of momentum, sensible heat, latent heat, and carbon dioxide. The fluxes are obtained with the eddy covariance technique, which involves correlation of the vertical wind component with the horizontal wind component, the air temperature, the water vapor density, and the CO2 concentration.
¶ Additional Information
Instruments used are
- A fast-response, three-dimensional (3-D) wind sensor (sonic anemometer) to obtain the orthogonal wind components and the speed of sound (SOS) (used to derive the air temperature) and
- An open-path infrared gas analyzer (IRGA) to obtain the water vapor density and the CO2 concentration.
The ECOR systems are deployed at the locations where other methods for surface flux measurements (e.g., energy balance Bowen ratio systems) are difficult to employ, primarily at the north edge of a field of crops.
The ECOR timestamp is at the beginning of the averaging period (i.e. 1600 covers 1600-1630), whereas the EBBR timestamp is at the end.
* For more information see ECOR
¶ Expected Operations
¶ Metrics
Normal Operations Metrics Table
The ECOR metrics should look mostly green as below. Some of the variables have nuisance flags that fail most of the time. These are nothing to be concerned with.
Rain/Dew Metrics Table
The ECOR metrics will look a little weird when there is precipitation or dew on the instrument. In this case, variables will be missing and many more will be flagged. This is normal for this instrument in these conditions.
¶ Plots
Note: The only co-located ECOR and EBBR is at E13/E14 (SGP Central Facility). Caution must be used in the comparison of the two systems because they usually see different vegetation surfaces. The best comparison can be made for straight north or northwest wind directions, when both systems view the same grass surface. For other directions, the two systems are viewing different vegetation surfaces and the fluxes from the two will probably not be similar, unless perhaps, the ground is snow covered. Except for the E14 and E21 sites, sensible and latent heat flux, and wind speed and direction measurements can be compared for adjacent ECOR and EBBR systems (which all view grassland), again remembering that there are likely to be climatologically driven differences. No other measurements can be reliably compared, mostly because the ECOR LI-7500 and sonic anemometer are not meant to produce accurate measurements of anything else that both systems measure.
Many of the plots will indicate when data are missing and/or failing QC. If data are missing for any reason (rain/dew/problems), then some of the plots will try and interpolate the data. This will be indicated by orange triangles at the top of the plot and orange lines for the interpolated data.
Heat Flux
The Sensible Heat Flux (h) and Latent Heat Flux (lv_e) should follow similar trends, only on different scales. When the EBBR data are plotted as a comparison, the Net Radiation (q) will have generally have a positive trend on a sunny day, peaking around solar noon. The EBBR latent heat flux (e) should be similar to the ECOR latent heat flux (lv_e). The EBBR sensible heat flux (h) should trend similar to the ECOR. Since they are 2 different instruments, the values may not be similar.
CO2
This shows the CO2 flux and mean CO2 density. The CO2 flux should follow a similar trend as the sensible heat flux.
Momentum and Friction
This shows the momentum flux (k) and the friction velocity (u*). You may be wondering what exactly momentum flux and frictional velocity physically represent. The former describes horizontal motion being translated up or down by vertical turbulence. Whereas, u* can be thought as a surface velocity scaling term which can be used in various ways (i.e., when making a log wind profile).
Occasionally, the momentum flux will be positive tripping the Failing QC flag. The positive momentum fluxes, at night especially and sometimes during the day, is an indication that stable conditions are occurring during the time period.
Meteorological Variables
Four of the plots show comparisons of the ECOR surface meteorology data and co-located data from similar instruments. These include:
Pressure
Temperature
Water Vapor
Wind
Wind Components
This plot shows the different components of the measured wind (u,v,w):
The MET and ECOR wind directions/speeds being offset is easily explained. The MET system timestamp (and the end of the averaging period) is the end of the half hour. The ECOR is measuring the same half hour average, but its timestamp is the beginning of the half hour. Therefore, the MET data point just to the right of the ECOR data point on a DQ-Explorer plot is from the same half hour.
Heats
This plot shows the specific heat of moist air (cp) and the latent heat of vaporization (lv). The specific heat values will have a step pattern to them. The latent heat should follow the trend similar to the moist air density in the plot below.
Moisture
This plot shows the mixing ratio (mr) and moist air density (rho). Note that rho may breach the upper limit of 1.35 kgm^-3 at northern sites during the winter (i.e., NSA and OLI) tripping the failing QC flag. If ever uncertain, it is always easy to do a quick back of the envelope calculation using the ideal gas law and ECOR pressure and temperature variables to confirm that the moist air density is reasonable.
¶ Site-Specific Information
¶ SGP
E21 - NOTE: site no longer in service
The plots of data from the forest site at Okmulgee show more “jumping around” of the data than is seen at the other ECOR sites; this is expected and normal since the scale of eddies that carry the flux information over the tree structure is much larger than over grassland or crops.
Fluxes of CO2, sensible heat, and latent heat at E21 Okmulgee forest are often larger than at other sites, particularly the fluxes of water vapor and CO2; the latter will often be twice what it is at the other ECOR sites.
The plots of data from the forest site at Okmulgee exhibit more noise than is seen at the other ECOR sites. This is expected and normal since the the tree structure presents a much rougher and less homogeneous surface than exists for grassland or crops.
All EFs
Wind Direction Dependencies (numbers are wind direction in degrees); these values indicate wind directions where the data is good. Note that there are wind directions at most sites for which the fetch is insufficient and therefore ECOR data is invalid. Changes in surface vegetation type and state can occur with time (refer to the EFSCO).
¶ AMF1 and AMF2:
Point Reyes, CA (PYE): For some wind directions, the horizontal fetch was not representative of the field in which the AMF was located. Therefore, for the wind direction ranges 66–92 (buildings and trees) degrees, the fluxes are affected by insufficient fetch and surfaces, buildings, or vegetation that are not similar to the local field conditions.
Niamey, Niger (NIM): For some wind directions, the horizontal fetch was not representative of the field in which the AMF was located. Therefore, for the wind direction ranges 90–170 (buildings) and 220–280 (trees) degrees, the fluxes are affected by insufficient fetch and surfaces, buildings, or vegetation that are not similar to the local field conditions.
Hesselbach, Germany (FKB): For some wind directions, the horizontal fetch was not representative of the field in which the AMF was located. Therefore, for the wind direction ranges 40–159 and 176–209 degrees, the fluxes are affected by insufficient fetch and surfaces, buildings, and vegetation that are not similar to the local field conditions.
Steamboat Springs, CO (STORMVEX, SBS): snow all directions
Shouxian, China (HFE): ungrazed grass all directions
Gracious Island, Azores, Portugal (GRW): Grass 100-260; grass and low shrubs 270-360 and 0-99
Cape Cod (TCAP, PVC): Seashore grass and shrubs all directions; some saltwater sea influence 0-100
Brazil (MAO): ungrazed grass all directions
AWARE (Antarctic, AWR): – West Antarctic Ice Shelf (WAIS): all directions snow/ice – McMurdo?: For some wind directions, the horizontal fetch was not representative of the tundra field in which the AMF2 was located. Therefore, the fluxes are affected by insufficient fetch and surfaces, buildings, and vegetation that are not similar to the tundra surface.
AMF3/OLI:
All directions tundra; all directions the fetch is limited
SGP:
EF1: 0–53, 120–360 wheat or wheat stubble
EF3: 0–48 pasture; 132–260 soybeans, wheat
EF5: 80–154 sorghum or wheat; 155–260 wheat or wheat stubble
EF6: 0–90 grazed pasture; 91–360 alfalfa and brome grass
EF10: 0–90, 270–360 grazed; 91–269 grass
EF14: 129–265 wheat (2004), corn (2005), soybeans (2006); but normally wheat; 352–85 ungrazed grass
EF16: 134–269 pasture; 334–360 ungrazed grass
EF21: 0–360 mixed deciduous forest (note that for the direction of the tower, 0–30, the data may be suspect)
EF24: 80–280 wheat or wheat stubble
EF31: 100-200 corn/soybean; pasture 30-80
EF33: 100-300 wheat; 40-80 grass
EF37: 135-260 wheat; 280-310 pasture
EF38: 150-260 wheat
EF39: 100-260 wheat; 280-360 and 0-80 ungrazed grass
EF41: 100-260 wheat; 280-360 and 0-80 ungrazed grass
ENA:
All directions ungrazed grass; all directions the fetch is limited
NSA:
EF10: All directions tundra; 340 through 20 the fetch is limited
EF11: 350-360 and 0-100 saltwater sea; All other directions beach gravel
¶ Known Behaviors
¶ Known Issues
Known issues that may not need to be mentioned in DQAs:
Notes
- Sudden shifts in wind direction are not handled well by the ECOR coordinate transform routine, often resulting in a spike in the fluxes ecfor a half hour measurement period.
- When pressure and temperature missing, default values are used in the calculation of the CO2 and latent heat fluxes. When default values are used, resulting errors in the fluxes should be within the ± 10% system error.
- Missing data periods occur at times. This is usually a site data system collection/communication problem. It can also occur if the ECOR data acquisition computer fails. Missing data is sometimes filled in later from manual or automatic re-collection of the data.
- The ECOR data file contains both the wind speed and direction, and the u,v,w components. Typical academic standard is for a positive U component to represent wind blowing to the East and positive V to the North. Unfortunately the ECOR U and V directions are shifted. Positive U component represents wind blowing to the North and positive V represents wind blowing to the West.
- Friction velocity (ustar) and momentum flux (k) are often flagged during light wind conditions, especially at the E21 Okmulgee forested site. This is normal, as these measurements, as well as the fluxes of sensible heat flux (h), latent heat flux (lv_e), and CO2 flux (fc) cannot be trusted because of the lack of ability of the sonic anemometer to measure properly during very low wind speeds (especially < 1 m/s)
- The moist air density occasionally exceeds 1.35kgm^-3 tripping the failing QC flag, especially at the NSA and OLI sites during the winter. Although this should be no source of concern, a quick back of the envelope calculation using the ideal gas law and ECOR variables should be able to confirm the density within reason.
- Periods of precipitation, fog, and dew (frost) often cause incorrect water vapor (mean_q), latent heat flux (lv_e), and carbon dioxide measurements (fc & mean_c). This is caused by water lying on the lower window of the LI-7500 sensor, thereby obstructing the passage of the sensing IR radiation (very light precipitation may have little or no effect). The CO2 portion of the instrument is more sensitive to this condition, so it is not unusual for latent heat flux (lv_e) to be good, even though the CO2 flux is not.
- IRGA temperature is the temperature inside the IRGA which is metal. Therefore, the IRGA temperature will always be higher than the EBBR or MET.
- The Sonic temperature is measuring virtual temperature and can be either greater or lower than the ambient temperature.
- IRGA water vapor density is very inaccurate. Do not bother noting unless the vapor density difference is greater than a factor of 3.
- Because the variance in pressure of the MET sensor is smaller than the resolution of the IRGA, changes in pressure can be so small that it appears the pressure is flatlining. This is not a cause for concern. See image below:
Quirks
The variance will fail if it is too small. This (or any flux related value) failure is an indicator that the flux value is not trustworthy. So, this is not really a bad thing, but could be just an indication that at that site there is little variance (a common condition for winter for carbon dioxide).
The Sensible Heat Flux (h) and Latent Heat Flux (lv_e) will not be directly equal, but should have the same sign of change, and sometime similar values. At the Central Facility the co-located EBBR is also plotted with the right axis reversed for comparison. The values for the EBBR are negative in value as compared to the positive values for the ECOR and the axis has been reversed to allow the direct comparison of EBBR Sensible Heat Flux (EBBR h) and Latent Heat Flux (EBBR e)
Report when RH is above maximum. This usually means the sensor has become dirty and needs to be replaced.
¶ Past Problems
Past problems that do need to be mentioned in DQAs:
Spikes in CO2 Flux
Spikes in the CO2 flux were observed at the AMF site in Graciosa Island, Azores. These spikes occurred around the same time (~16:30 GMT) for a period of days in which the winds were out of the East.
The site is located next to an airport. A plane would land around the same time each day and a diesel generator would be hooked up to the plane to act as an auxiliary power unit. Since the winds were out of the East, the exhaust was measured by the ECOR, thus creating a spike.
Missing Data
A common problem with the ECOR systems that may results in loss of data for the whole datastream is a failure of the RocketPort Hub or the Sonic anemometer. This problem has been seen numerous times and is documented in a number of DQPRs: 263, 301, 320, 330, 372, 375
Failed Licor (aka IRGA)
The Licor/IRGA collects data for the atm_pres, fc, lv_e, mean_c, mean_q, temp_irga. When all of these variables fail to produce data at the same time, a likely problem is a failed Licor or the connection between the Licor and the RocketPort Hub. See DQPR 396.
Instrument Time Not Correct
After a PM the instrument time was 7.5 hours off and behind the co-located SMOS. This caused the two not correctly match when plotted together. Notice how the first plot shows a large jump in temperature around 08:00 GMT, and the second plot shows the day's high temperature to be at sunrise. Also, the ECOR Sonic temperature profile matches the SMOS temperature very well, but is shifted in time.
Also note that the reporting of the ECOR data is at the beginning of the half hour, whereas the EBBR and SMOS report at the end of the half hour. So a peak in an ECOR measurement will show on the plots as being half an hour before the peak in the EBBR or SMOS. See DQPR 602
Garbled Data
The ECOR at can occasionally enter the "garbled" data condition, where the sonic datastream length is not always normal and the ECOR program can not properly interpret the data. To fix this it is necessary to restart the ECOR program until the condition goes away. See DQPR 2270