1 / 39

Uncertainties in Eddy Correlation Techniques: Data Types and Calibration Strategies

Learn about the challenges and strategies for calibrating and analyzing data in Eddy Correlation techniques, focusing on vegetation height, data heterogeneity, and flux measurement uncertainties. Understand factors affecting accuracy, error sources, and implications for flux calculations.

landen
Télécharger la présentation

Uncertainties in Eddy Correlation Techniques: Data Types and Calibration Strategies

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. CAMELS- uncertainties in data Bart Kruijt, Isabel van den Wyngaert, Ronald Hutjes, Celso von Randow, Jan Elbers, Eddy Moors...

  2. Types of data vegetation height, LAI, d, z0, rooting … heterogeneity, sampling cup anemometer stalling, hygrometers.. calibration, dew on radiation sensor,.. Sheltering, shading, … =(w2c2) *T/ Tscale --> fourth moments calibration, pump maintenance, window cleaning averaging time, coordinate rotation, freq. corr footprint models, heterogeneity, win direction calm nights drainage, return fluxes • Land use, Site parameters • accuracy • representativity • driving variables (weather) • instrument error/precision • technical/ operational error • siting error • validation/optimisation data (fluxes) • stochastic error • technical/ operational error • calculation/conceptual uncertainty • representation of surface • day • night

  3. Fc =.w.c NEE = Fc + z(c/ t) Eddy correlation ? CO2

  4. Eddy correlation hopeless?

  5. Each processing step carries uncertainty

  6. Time Sensitivity to flux calculation methods Rotation: correction for tilt of mean streamlines Detrending and averaging: removing non-stationarity

  7. CO2 Fluxes (SW Amazon) - Scale contributions ‘Turbulent’ ‘Meso-scale’

  8. Summary effects of rotation and averaging Relative effects of averaging time and rotation on daily total fluxes, Amazon

  9. Longer averaging times --> better energy closure? Finnigan, Malhi, 2002

  10. Total uncertainty from rotation and averaging over the day

  11. Uncertainty in calibration Calibration a posteriori causes problems and uncertainty

  12. Eddy flux, storage flux and Ecosystem (‘biotic’) flux Windy nights Calm nights

  13. Eddy correlation integrates everything but misses advection Morning Rs CO2 return ? Rs Rs Night CO2 drainage ? Rs Manaus, Amazon CO2 stored in valleys

  14. Total one-sided error for AMAZON on annual totals is, apart from night-time error, between 12.5% and 32%, or 1-2 t ha-1.

  15. Systematic or random error? • Error depends on measuerement height, surface type, time of day, weather • Random error vanishes when the number of independent samples increases. • BUT: when are atmospheric samples independent? • Systematic error is persistent. • What if maintenance varies or calibration drifts? • What if low frequencies vary with weather or season? • ---> when do systematic errors become random?

  16. Bias. Example from the SW Amazon, with cold periods

  17. Other bias : transient periods (morning, early evening) are non-stationary and carry high uncertainty rainy periods carry high uncertainty ideal weather associated with specific wind directions

  18. Estimates for CAMELS

  19. Rebmann et al - CARBOEUROFLUX footprint-quality analysis

  20. Discussion: • How to avoid bias when applying uncertainties to model fitting? Include more processes? Look at daily totals where day-night cross contamination occurs? • Can we eliminate bias by better matching models and measurements? • How to fine-tune uncertainties for specific sites or conditions?

  21. Consider the area beneath the sensor a leaky, sloshing vessel and fit both physiological and micrometeorological parameters Fc=f(C,u*,lm,R,Ps) U* • lm C=sum(R-Ps-Fc-advection) Advection=f(C) Advection R, Ps=alpha.PAR To be tested ….

  22. Some early results look good

  23. Effect of spikes in one channel only 5 ppm and 50 ppm spike on CO2. Effect is random relative uncertainty, increasing with spike/signal ratio

  24. Uncertainty in tube delay calculations

  25. Summary effects of rotation and averaging Variation in sensitivities to treatments Relative effects of averaging time and rotation

  26. Frequency corrections Zero-plane, tube NOT important. Low frequencies ARE important.

  27. Conversion ppm m s-1 to area based fluxes Small potential errors average out over days

  28. Similarity relations - representativity for surface Filtering for poor similarity will discard important periods such as early morning

  29. Uncertainty as a function of the percentage good data - Rebio Jaru

  30. Uncertainty on annual totals from (well distributed) data gaps

  31. And finally….

More Related