Processed Eddy Correlation data location no. 16 in CASES99 network of flux towers version 1.0 1. Authors and data source: Wageningen University (the Netherlands) Oscar Hartogensis, Bas van de Wiel 2. Description of the sensors 2.1. Sensor Type Eddy correlation systems consisting of: a. Campbell Sci. CSAT3 sonic b. Campbell Sci. KH20 Krypton fast response hygrometer c. Campbell Sci. FW05 Thermocouple A Campbell 23X datalogger, which had two eddy correlation systems as described above (one at 10.2m and one at 2.65m) hooked up to it, registered the data which was subsequently automatically downloaded onto to a laptop computer. 2.2. Specifications: a. Campbell Sci. CSAT3 sonic: Path length over which the sensor integrates = 10 cm vertical, 5.8 cm horizontal Transducer Path Angle from Horizontal = 60 degrees Threshold = 30 mm/s for u and v; 4 mm/s for w b. Campbell Sci. KH20 Krypton fast response hygrometer Path length over which the sensor integrates = 1.4 cm Response time = 0.01s c. Campbell Sci. FW05 Thermocouple Diameter = 0.0005 inch Type = chromal – constantan Response time = better than 0.05s 2.3. Location: Location no. 16 in CASES99 network of flux towers; co-ordinates N37’38.611’ W096’44.233’ 2.4. Set-up: 10m System: Height above the ground = 10.20m distance (horizontal) between Sonic and Krypton = 0.13m distance (horizontal) between Sonic and Thermocouple = 0.05m Directional off-set relative to North (0 deg) of Sonic = 65 degrees (+/- 5 degrees) Path length of Krypton = 0.014m 3m System: Height above the ground = 2.65m distance (horizontal) between Sonic and Krypton = 0.11m distance (horizontal) between Sonic and Thermocouple = 0.05m Directional off-set relative to North (0 deg) of Sonic = 50 degrees (+/- 5 degrees) Path length of Krypton = 0.014m 2.5. Maintenance: The system was stopped during maintenance work on the tower. Every day the Krypton was cleaned with a moist Q-tip, thermocouple was checked and spider webs were removed. Once every two or three days the datalogger time was set to the GPS based time of the NCAR system within an accuracy of 1 second. 3. Description of derived variables Raw 20Hz eddy correlation data of both the 3m and 10m systems were processed into 5 minute and 30 minute fluxes using eddy correlation processing software developed at Wageningen University. Apart from calculating straightforward means, standard deviations and covariances some corrections are performed on the data which will be briefly described here: - The Krypton humidity sensor is base lined with accurate slow humidity measurements - The Krypton hygrometer measurements are corrected for it’s sensitivity for oxygen; the sensors used during CASES99 were individually calibrated to determine their oxygen sensitivity; these were found to be the same: the extinction coefficient for oxygen found is 0.0038 - The sonic temperature is corrected for humidity influence on the measurement - For every individual averaging interval (5min or 30min) the sonic coordinates are rotated such that mean(w)=0, mean(v)=0 and cov(v,w)=0 - NO frequency response corrections were applied - The 30 minute averaged data was linearly detrended before standard deviations and covariances were calculated. - The 5 minute averaged data was NOT detrended. - Structure parameters of temperature, humidity and correlated temperaure-humdity were calculated using Taylor’s frozen turbulence hypothesis: eg. CT2 = avg( (T2 – T1)^2 / (U*t)^2/3 ) where CT2 is the structure parameter of Temperature T is temperature U is the mean wind vector t is the time-interval between the measurements of T1 and T2 The delay factor U*t replaces the separation distance R in the traditional definition of the structure parameter. Here R is chosen to be as close as possible to 15cm. Atmospheric pressure needed to calculate the fluxes was taken from the ISS towers 5 and 6, which were the two nearest ISS towers to location 16. There are four eddy correlation flux files: 10m system at 30 minute intervals 3m system at 30 minute intervals 10m system at 5 minute intervals 3m system at 5 minute intervals The files are in ASCII format. The data columns are separated by a semi-colon (;). Invalid or no data are given as a dummy “-9999”. Data is rejected by the processing software when: - The CSAT3 sonic diagnostic parameter is set to an error value - The raw data is checked for it’s physical meaningfulness (air temperature below 70 degrees, no negative humidity etc.); with this check also dummy values in the raw data are filtered out. - When less than 80% of the expected data samples are present in a time interval The resulting standard deviations, fluxes and structure parameters have not been filtered further in any way. The columns in the data files are labelled; here follows an overview of all the variables with their label names and units: YYYYMMDDHHMMSS : Time in CASES99 standard time format [UTC] DOY : Day of Year [CDS] Hr : Hour at end of 5 or 30 min interval [CDS] Mn : Minute at en d of 5 or 30 min interval [CDS] #samples : total number of datalines found in 5 or 30 min time interval #U : number of error free wind speed in x-coordinate samples found in 5 or 30 min time interval #V : number of error free wind speed in y-coordinate samples found in 5 or 30 min time interval #W : number of error free wind speed in z-coordinate samples found in 5 or 30 min time interval #TSon : number of error free sonic temperature samples found in 5 or 30 min time interval #TCop : number of error free thermocouple samples found in 5 or 30 min time interval #RhoV : number of error free hygrometer (Krypton) samples found in 5 or 30 min time interval #q : number of error free hygrometer (Krypton) samples found in 5 or 30 min time interval #time : number of valid time indicator samples found in 5 or 30 min time interval Dir : wind direction corrected for off-set relative to North in the set-up [deg] d(dir) : tolerance in Dir [deg] Mean(U) : mean wind speed (vector between u and v) [m/s] dMean(U) : tolerance in Mean(U)[m/s] Mean(TSon) : mean sonic temperature [K] dMean(TSon) : tolerance in Mean(TSon) [K] Mean(TCop) : mean thermocouple temperature[K] dMean(TCop) : tolerance in Mean(TCop) [K] Mean(q) : mean humidity mixing ratio[kg/kg] dMean(q) : tolerance in Mean(q) [kg/kg] sigma(TSon) : standard deviation of sonic temperature[K] dsigma(TSon) : tolerance in sigma(TSon) [K] sigma(TCop) : standard deviation of thermocouple temperature[K] dsigma(TCop) : tolerance in sigma(TCop) [K] sigma(q) : standard deviation of humidity mixing ratio[kg/kg] dsigma(q) : tolerance in sigma(q) [kg/kg] sigma(u) : standard deviation of wind speed in x-coordinate[m/s] dsigma(u) : tolerance in sigma(u) [m/s] sigma(v) : standard deviation of wind speed in y-coordinate [m/s] dsigma(v) : tolerance in sigma(v) [m/s] sigma(w) : standard deviation of wind speed in z-coordinate [m/s] dsigma(w) : tolerance in sigma(w) [m/s] cov(TSon,q) : covariance between sonic temperature and mixing ratio [K.kg/kg] dcov(TSon,q) : tolerance in cov(TSon,q) [K.kg/kg] cov(TCop,q) : covariance between thermocouple temperature and mixing ratio [K.kg/kg] dcov(TCop,q) : tolerance in cov(TCop,q) [K.kg/kg] cov(TSon,U) : covariance between sonic temperature and mean wind vector [K.m/s] dcov(TSon,U) : tolerance in cov(TSon,U) [K.m/s] cov(TCop,U) : covariance between thermocouple temperature and mean wind vector [K.m/s] dcov(TCop,U) : tolerance in cov(TCop,U) [K.m/s] cov(q,U) : covariance between mixing ratio and mean wind vector [kg/kg.m/s] dcov(q,U) : tolerance in cov(q,U) [kg/kg.m/s] H(Sonic) : sensible heat flux using sonic temperature[W/m2] Tol(HSonic) : tolerance in H(Sonic) [W/m2] H(TCouple) : sensible heat flux using thermocouple temperature [W/m2] Tol(HTCouple) : tolerance in H(TCouple) [W/m2] LvECov : plain latent heatflux [W/m2] dLvECov : tolerance in LvECov [W/m2] LvEWebb : Webb-term of latent heat [W/m2] dLvEWebb : tolerance in LvEWebb [W/m2] LvE : total latent heatflux (LvECov + LvEWebb) [W/m2] dLvE : tolerance in LvE [W/m2] UStar : U* [m/s] dUStar : tolerance in UStar [m/s] Tau : shear stress[N/m2] dTau : tolerance in Tau [N/m2] R(delay) : separation distance for current 5 or 30 min interval for structure parameter calculations [m] dR : average separation distance (U*t) after shifting 1 sample only; R(delay) will made up of a multiple of dR to get as close as possible to the R(delay) chosen [m] CTSon2 : structure parameter of sonic temperature [K^2/m^2/3] dCTSon2 : tolerance in CTSon2 [K^2/m^2/3] CTCop2 : structure parameter of thermocouple temperature [K^2/m^2/3] dCTCop2 : tolerance in CTCop2 [K^2/m^2/3] Cq2 : structure parameter of mixing ratio [kg/kg.m^-2/3] dCq2 : tolerance in Cq2 [kg/kg.m^-2/3] CTSonq : structure parameter of sonic temperature and mixing ratio [K.kg/kg/m^2/3] dCTSonq : tolerance in CTSonq [K.kg/kg/m^2/3] CTCopq : structure parameter of thermocouple temperature and mixing ratio [K.kg/kg/m^2/3] dCTCopq : tolerance in CTCopq [K.kg/kg/m^2/3] 4. Sample period Sample period = 0.05s (20Hz) Averaging period = 5 minutes or 30 minutes From 30 Sept. until 20 Oct. the datalog system suffered overrun problems in writing the data to the PC; this means that effectively the system was not running on 20Hz; missed data lines can be detected in the diagnostic counter or the variable “Sec” (seconds) which skip a value from one line to the next when a data line was missed. From 1020 onward this problem was solved at the cost of not registering the thermocouple data of the 10m eddy correlation system. Also the “raw” (no corrections, averaging preformed) eddy correlation data are available in NetCDF. 5. Data availability Eddy correlation data are available between 0930 18:30 and 1028 9:00 local (CDS) time. Daily the system was switched of for maintenace as well during work on other instruments in the tower. Therefore during the day there some gaps in the data. However, all the night data is complete. As mentioned in part 4 from 1020 onward no thermocouple data of the 10m eddy correlation system were available.