Hallo,

the  resolution of 0.01 m/s is that velocity which can be detected. The accuracy depends on the absolute value and is 1.5% (on avarage). This is e.g. 0.15m/s if the wind speed is 10m/s. 

If you have a measured time series, the measurement accuracy is not known at a particular time. The statistical distribution of the "measurement noise" is known only. The measurement uncertainty or noise is thus dominated by resolution for "no wind" situations. and by measurement accuracy otherwise. I hope, it helped.

Thank you very much Andreas. It is immensely helpful. One more thing which I would like to ask, why for the sonic measurements the RMS noise is given as accuracy but for the gas measurements it is given as precision?

One clarification: 

"0.15m/s if the wind speed is 10m/s" 

If I have a time series of 30 min wind data, then windspeed of 10m/s would mean the averaged value over 30 minutes right?

Is it as per manufacturer's convenience because as far as I know, accuracy and precision are different terms.

The overall errors in flux measurements by Eddy Covariance is about 15% for CO2 flux, about 5-10% for H2O flux and about 5% for sensible heat flux when the setup is very clean and the processing is full and complete.

But is will still vary with instrument height, position of instruments vs each other and wind direction (sonic anemometer shaded or blocked by analyzer), and structure of turbulence for each 30 minutes.

So, overall error is comprised of instrument performance, setup, processing and ambient conditions.

Dear Subharthi,

during my studies I worked on particle eddy covariance measurements. During that time the book "Micrometeorology" by Foken (2008) was very helpful for designing an Eddy Covariance System and to address the uncertainties which consists obviously not only of instrument related accuracies. Do you know this book?

Best regards,

Florian  

Dear Dr Burba and Florian,

Thank you very much for answering the query. It is really very helpful and I very much appreciate that.

Re Dr Florian: I know about the book by Foken (2008)  and will surely have a look at it.

Thanks and Regards

Subharthi

Suggest you read:

Rowe, M.D., C.W. Fairall, and J.A. Perlinger, 2011: Chemical sensor resolution requirements for near-surface measurements of turbulent fluxes.  Atmos. Chem. Phys., 11, 5263–5275, doi:10.5194/acp-11-5263-2011.

Blomquist, B. W., L. Bariteau, J. W. Edson, C. W. Fairall, J. E. Hare, W. R. McGillis, B. J. Huebert, S. D. Miller, and E. S. Saltzman, 2014: Advances in ship-based air-sea CO2 flux measurement by eddy covariance.  Bound.-Layer Meteorol., 152, 245–276, DOI 10.1007/s10546-014-9926-2.

Dear Dr Fairall,

Thank you very much  for your kind help. I will surely look into the above references. 

Thanks and Regards

Subharthi

The central limit theorem (CLT) tell us that the static noise, reported as standard deviation, should decrease with the square root of integration time. This is only theoretical as drift counteracts this trend. In order to have proper information, precision should be reported not just as a figure, but as function of integration time (e.g. as the Allan variance) or at least as precision for a given integration time. We have inquired but have been unable so far to obtain this information for our Gill sonic anemometer. Let me thus add to your question: what I mean is that I don't know if your Gill's coeficients of variation are for 1 min or for 0.1 s integration times. Maybe GEORGE BURBA and others could help us with this.

Dear Dr Rodrigo

Thank you very much for the clarification. Even I would like to hear what Dr Burba has to add in this.

Thanks and Regards

Subharthi

The white noise level of the Gill can be determined by plotting the variance spectrum vs frequency.  At higher frequency it may flatline at the noise level.

Note, the turbulence signal is often well above the Gill level even at 20 Ha, so you need to find a light wind period.  The characterization the white noise in (m/s)^2/hz is the best approach.

cf

Thanks much, Chris.

Thank you very Dr Fairall