Quality Assurance
and Control

Summary

Since August 2002, the FS portable EC system from Durham, NH has been used for site intercomparisons. Setup, maintenance and calibrations followed those that had been used previously. Table 1 shows a list of sites visited and status of the intercomparison. We have replaced the old portable system because of data quality issues. An assessment of data quality of the FS EC system used in 2000-2003 is posted on the AmeriFlux web page (http://public.ornl.gov/ameriflux/standards_roving.shtml).

Highlights

• Temperature. All sites visited used HMP Viasala shielded (not aspirated) sensors. All but one site had higher daytime air temperatures compared to our standard (aspirated). Offsets ranged from + 0.25 to 1.1 ºC. Higher (daytime) temperatures are likely a function of the type of radiation shielding. No site visited had calibrated their temperature sensor.
• Net Radiation. Estimates from our REBS Q*7.1 sensor compared very well to other Q*7 sensors, regression slopes ~ 1 and offset range -3 to 8 W m-2. Comparisons of our Q*7 data to sites that use the Kipp and Zonen CNR-1 radiometer, show the Q*7 underestimates net radiation by 12 - 123 W m-2. The CNR-1 measures a larger frequency range and is used by ~ 37% of AmeriFlux sites, so we now have CRN-1s for comparisons at these sites.
  
• PPFD. All sites visited used the li-190 series sensors for PPFD estimates. If sensors are well maintained and periodically calibrated, they compare well (regression slopes ± 2 %, offsets ± 10 mol m-2 s-1). However, 2 sites that did not maintain their sensors showed lower PPFD by ~ 20 %.
  
• FCO2. Carbon fluxes that were not corrected for high frequency losses compared well, particularly for nighttime fluxes (Figure 1A). At one site, comparison with another li-6262 using shorter tubing lengths and a higher flow rate suggested that our old EC system underestimated exchanges by ~ 23 %. When our 6262 data were compared with data from an open-path sensor (operating at higher frequency with little or no sensor separation), we underestimated fluxes by ~ 68 % (Figure 1B). Our new hardware and the ability to reduce sensor separation facilitates high frequency measurements and the new software will help us to better assess the high frequency components of turbulent exchange in a timely manner.
  
• H. Measured fluxes ranged ± 20% with minimal offsets < 5 W m-2. No systematic reasons could be found.
  
• AE. Sites with similar sensor separation distances but with dirty or clogged tubing showed a 14 - 30 % reduction in E fluxes (with same flow rates and tubing lengths). Our measurements were 5 - 12 % lower compared to sites using higher flow rates and shorter tubing lengths.