To: russell.monson@colorado.edu, laura.scott@colorado.edu, jia.hu@colorado.edu, nicole.trahan@colorado.edu, lynette.laffea@colorado.edu, Diego.Riveros-iregui@colorado.edu Subject: Updates to Monson/AmeriFlux Data.... Date: Thu, 05 Mar 2009 14:28:36 -0700 From: "Sean Burns" Hi All, Attached below is an update about the Monson Niwot Ridge Ameriflux data...if you want to be removed from this email list just let me know and i'll take you off....i hope to have a "final" 2008 data and updates to previous years available sometime in the May/June time frame.... thanks, SpB. - ------------------------------------------------------------------------------ Sean Burns Department of Ecology and Evolutionary Biology (EEB) Campus Box 334 University of Colorado Boulder, CO 80309-0334 for FEDex: Ramaley Building, Room N122 Internet: sean.burns@colorado.edu http://www.mmm.ucar.edu/people/burns/ Phone: (303) 492-5796 Fax: (303) 492-8699 - ------------------------------------------------------------------------------ Updates to report about the CU/Monson AmeriFlux data. ...[stuff deleted]... 4. In looking over the "ustar correction" from past years, there appears to be some inconsistencies. I won't get into whether or not the ustar correction is appropriate or not, because this is a separate issue (eg, see Acevedo, et al 2009 for a discussion about this). For the years 2004-2007, the ustar-correction was not applied in the winter....it's windy in the winter so ustar is typically not low, but this still makes a small difference. For the 2002 data, the coefficients used to do the ustar-correction were unusually large..this made the ustar-corrected data too large and had a significant impact the NEE for that year....SO, to make the data more consistent from year-to-year we decided to use a simple scheme for the correction...one set of coefficients in the winter and a different set in the summer. What we are using is: Equation: F_co2= A*exp(B*T.soil) For fall/winter (days 1-140 and 273-366), A=1.07; B=0.108; For spring/summer (days 141-272), A=1.35; B=0.086; A more rigorous method such as that by Gu, et al, 2005 could be used and we might change to that at a later date (of course anyone can do this themselves, if they are so-inclined). For now, this is an improvement over what currently exists. And i believe that Bai Yang and Dario Papale, etc are going to do their own ustar-correction-method for the "level4" data files available through the ameriflux webpage, ie: ftp://cdiac.ornl.gov/pub/ameriflux/data/Level3/Sites_ByName/Niwot_Ridge/ ...Because we are going to update the older flux data I was also planning to take this opportunity to clean up some of the spikes which exist in the older data following procedures in Papale D, et al, 2006 or something similar to that. However, the effect of the "despiking" should be somewhat small....so that will not be implemented until the final data release. For anyone that is actively working with co2 NEE data I have updated data that use a consistent ustar filter for each year on the web at: http://urquell.colorado.edu/data_ameriflux_test/ the files are: - -rw-r--r-- 1 sburns 647028 Mar 5 09:58 niwot_NEE_fixed_1999.csv - -rw-r--r-- 1 sburns 649481 Mar 5 09:59 niwot_NEE_fixed_2000.csv - -rw-r--r-- 1 sburns 648423 Mar 5 09:59 niwot_NEE_fixed_2001.csv - -rw-r--r-- 1 sburns 629704 Mar 5 10:00 niwot_NEE_fixed_2002.csv - -rw-r--r-- 1 sburns 629782 Mar 5 10:00 niwot_NEE_fixed_2003.csv - -rw-r--r-- 1 sburns 630058 Mar 5 10:00 niwot_NEE_fixed_2004.csv - -rw-r--r-- 1 sburns 627483 Mar 5 10:01 niwot_NEE_fixed_2005.csv - -rw-r--r-- 1 sburns 696531 Mar 5 10:01 niwot_NEE_fixed_2006.csv - -rw-r--r-- 1 sburns 696557 Mar 5 10:02 niwot_NEE_fixed_2007.csv the columns in each file are: where the columns are: day s_Fco2_21m_nee year mo day hr min sec of year no ustar with ustar 2007, 1, 1, 4, 45, 0, 1.1979, 0.97524, 0.97524 2007, 1, 1, 5, 15, 0, 1.2188, 0.48794, 0.48794 2007, 1, 1, 5, 45, 0, 1.2396, 0.8703, 0.8703 2007, 1, 1, 6, 15, 0, 1.2604, -0.19222, -0.19222 so column 8 should correspond to what you have in your current data file (as the non-ustar corrected co2 "nee" flux)....and column 9 is the "new" ustar-filtered co2_nee data. If you are not actively working on these data I would suggest that you wait until the final files are released....if you are working actively on something it may be a good idea to check and see how these changes affect your results. And here are the values for each year: NEE [gC m-2], (summed from 1 Jan - 31 Dec) year no_ustar ustar (new) ustar (old) effect of ustar filter - ----- -------- ---------- -------- --------------------- 1999 -160.73 -87.95 -88.48 72.78 2000 -117.76 -48.60 -48.83 69.16 2001 -142.76 -78.30 -82.95 64.46 2002 -107.95 -41.66 -20.73 66.29 2003 -95.89 -52.94 -67.10 42.95 2004 -91.43 -57.38 -62.20 34.05 2005 -112.66 -69.41 -88.01 43.25 2006 -129.49 -95.35 -104.36 34.14 2007 -123.13 -89.28 -97.85 33.85 2008 -120.53 -97.66 -97.66 22.87 the thing that i'm a bit puzzled over is why the effect of the ustar filter was realively larger in early years compared to the past several years...something subtle is probably happening here. For anyone interested, here are more plots and discussion about this: - ----------------------Further details/plots about ustar stuff: The main plot about the ustar-filtering is: http://urquell.colorado.edu/sean/plots_interyearly/plot_ustarfilter.html the upper row of 3-panel plots are: - difference of ustar-filtered NEE and non-filtered - cumulative nee over the year - the difference between ustar-filtered NEE - non-filtered (for both the "old" and "new" fit-coefficients. the lower row of 2-panel plots are related to the equation, F_co2 = A*exp(B*T.soil) and are: - the "A" coefficient - the "B" coefficient for the years 1999-2002 the difference between ustar-filtered and non-filtered is not zero for ustar > 0.2. I'm not sure why this is happening, but it will require re-processing the raw hi-rate data which is a long-term goal, but is not going to happen before next year. Also, a plot of the A/B data for each year is: http://urquell.colorado.edu/sean/plots_interyearly/plot_A_B_allyears.pdf this is probably the plot that shows most clearly the changes in the A/B values with time. Note that the black lines are what was used previously (see high A value for 2002), the blue lines are from the "sliding window" fit to the data, and the red lines are what we are using as the "new" coefficients. These also seems to be a distinct change in the sliding window values for the years 2006-2008 compared to earlier years....this coincides with changing our "reference" soil temperature sensor...but I'm not sure if it's the reason for this change... References: - ----------- Acevedo OC, Moraes OLL, Degrazia GA, et al. 2009: Is friction velocity the most appropriate scale for correcting nocturnal carbon dioxide fluxes?, Agricultural And Forest Meteorology, 149, 1-10. Gu, et al, 2005: Objective threshold determination for nighttime eddy flux filtering, Agricultural And Forest Meteorology, 128, 179-197. Papale D, Reichstein M, Aubinet M, et al. 2006: Towards a standardized processing of Net Ecosystem Exchange measured with eddy covariance technique: algorithms and uncertainty estimation, BIOGEOSCIENCES, 3, 571-583.