干旱区生态环境知识资源中心

The eddy covariance (EC) technique measuring the turbulent exchanges of heat, moisture, CO2 and momentum between surface and atmosphere has been used widely. However, the use of EC is based on some assumptions which do not exist in practice. So the results are not accurate without necessary corrections. Using data collected at the Semi-Arid Climate and Environment Observatory of Lanzhou University (SACOL), this work studies a method of acquiring high quality surface turbulent flux. In this approach, eddy covariance data processing includes eliminating despikes, coordinate rotations, sonic temperature correction, and WPL correction (correction for density fluctuations). Double rotation and planar fit are used to do coordinate rotations. The fluctuations of sonic temperature include the effect of humidity on the speed of sound, and should be converted into actual temperature. WPL correction was required to latent heat flux and CO2 flux for the density effects due to heat and water vapor transfer on turbulent flux measurements. Quality control is performed by the stationarity test, integral turbulent characteristics (ITC) test and overall quality is controlled to permit selecting high quality data. The results show that sonic temperature correction decreases sensible heat flux by about 7. 3% when using DR, but 5. 9% when using PF. The stability can influence the results of sonic temperature correction. WPL increases latent heat flux by 7. 4% and decreases CO2 flux by 72. 2%, respectively. Three coordinate rotations have great influence on momentum but little on scalar fluxes. The values of u(*) obtained from DR and PF are decreased by 6% and 3%, respectively. Only considering the dominant wind direction, the wind direction has no relationship with the correction of u(*) by DR which reduces u(*) by 5%. As the turbulent exchange strengthened (u(*) > 0. 3 m.s(-1)), PF and FPF gradually exhibit that the corrected values in southeast are bigger than the northwest area, PF in the southeast wind area increases u(*) by 9. 23%, yet decreases 3. 86% in the northwest area and FPF in southeast wind area increases u(*) by 10. 09%, yet decreases 1. 18% in the northwest area. The differences of the steady state tests between DR and PF are mainly in A parameterization scheme of DR and PF for SACOL provided in the ITC test works well. The overall quality shows that about 45%similar to 62% of the total data is higher for u(*) 66%similar to 68% for sensible heat flux, 62%similar to 65% for latent flux and 52%similar to 54% for CO2 flux. The proportion of the high quality of u(*) obtained by DR is 17% higher than PF, while that of the latter three kinds of fluxes obtained by PF is 2% 3% higher than DR. The difference between PF and FPF is mainly in u(*). Comparing the three coordinate rotations in the dominant wind direction, DR still obtains the best quality of u(*). The use of DR is recommended in the complex terrain for reducing calculation and improving the data quality.