Level-1: Final flux calculations

Level-1: Final flux calculations#

Info

Level-1 are the final flux calculations. Results from Level-1 are used in all subsequent processing steps.

Download EddyPro settings and metadata files

All used EddyPro processing settings and metadata can be downloaded directly from the repository here.

fluxrun

fluxrun v1.4.1 (using EddyPro v7.0.9) was used for all flux calculations.

Key settings for flux calculations#

Important settings

Sonic anemometer: NO angle-of-attack correction, see below
Time lag: Small time lag search window with nominal (default) lag for CO2 and LE from LI-7500
Time lag: Constant time lag for N2O and CH4 fluxes from QCL/LGR
Spectra and Cospectra QA/QC: low flux values for N2O and CH4, see below

EddyPro settings#

Project creation#

Raw file format: ASCII plain text
Metadata file: Use alternative file
Biomet data: Use external file
Metadata:
- Timestamp refers to: Beginning of dataset period
- File duration: 360 min
- Acquisition frequency: 20 Hz
- Canopy height: 0.50 m (constant across all years, dynamic canopy height was not available for most years)
- Displacement height: 0.40 m
- Roughness length: 0.08 m
- Altitude: 393 m
- Latitude: 47° 12’ 36.799’’ N
- Longitude: 008° 24’ 38.300’’ E
- Example for instrument settings from 2022:
  - Note that settings are different for some periods, but the settings listed here are representative examples used during flux processing.
  - Anemometer: Gill R3-50, Height: 2.41 m, Wind data format: U, V & W, North alignment: Axis, North off-set: 7.0°
  - Gas Analyzers:
    - LI-7500 IRGA: Northward separation: 4.00 cm, Eastward separation: 35.00 cm, Vertical separation: -1.00 cm
    - QCL (Aerodyne): Northward separation: 3 cm, Eastward separation: 0 cm, Vertical separation: -18.00 cm, Tube length: 857 cm, Tube inner diameter: 4.0 mm, Nominal tube flow rate: 8.0 l/m, Longitudinal path length: 32.00 cm, Transversal path length: 4.1500 cm, Time response: 0.1308 s
    - LGR (Los Gatos Laser, Generic Closed Path): Northward separation: 0 cm, Eastward separation: -5.00 cm, Vertical separation: -18.00 cm, Tube length: 954 cm, Tube inner diameter: 10.0 mm, Nominal tube flow rate: 8.0 l/m, Longitudinal path length: 28.00 cm, Transversal path length: 4.4450 cm, Time response: 0.1 s
- Raw File Description:
  - Original raw data files were converted from an irregular compressed binary format to a regular ASCII format using bico
  - Field separator: Comma
  - Number of header rows: 3
  - Columns were defined depending on the year or time period
  - ASCII files contained data for (in order): sonic anemometer, IRGA and QCL or LGR (if available)

Basic settings#

Files Info: this section in EddyPro was handled by fluxrun
Missing samples allowance: 10%
Flux averaging interval: 30 min
North reference: Use magnetic north
Master anemometer: R3-50
Gas measurements:
- CO2: using IRGA mole fraction for fluxes
- H2O: IRGA mole fraction (used as main flux), for some periods QCL/LGR mole fractions (only used as auxiliary flux)
- N2O: QCL/LGR mixing ratio
- CH4: QCL/LGR mixing ratio
Cell measurements:
- Average Cell Temperature: used for QCL/LGR
- Cell Pressure: used for QCL/LGR
Ambient measurements:
- used data from external biomet data file
Diagnostic measurements:
- Not directly used since the diagnostic value was only available for newer years. However, the AGC (automatic gain control, a measure of signal quality/strength) was used in all years.

Advanced settings#

Processing options#

Angle-of-attack correction: NO, because for another site we observed a tendency to e.g. overestimate CO₂ uptake, and to underestimate CO₂ release.
Axis rotations for tilt correction: Double rotation
Turbulent fluctuations:
- Detrend method: Block average
Time lag compensation:
- Settings for length of the lag search window, nominal (default) and constant time lags are based on reesults from Level-0 flux calculations, which uses a large time window for covariance maximization.
- IRGA: Covariance maximization with default, very small time window with default lag, similar to ICOS (see Table IRGA lags)
- QCL/LGR: constant time lag, separate for each year or for time periods when the time lag changed (see Table QCL lags)
Compensate density fluctuations (WPL terms):
- IRGA: YES because it is open-path
- QCL/LGR: NO because it is closed-path with an inlet tube and internally converts wet mole fraction to dry mole fraction, i.e., it directly records mixing ratios
Add instrument sensible heat components for LI-7500: NO (read more about this decision here)
Quality check: Mauder and Foken (2004) (0-1-2 system) for Level-1 quality flags, later extended in post-processing
Footprint estimation: Kljun et al. (2004)

Statistical analyses#

All tests were selected for output, but not all tests were later used. Notable difference to default settings: Angle of attack criteria were slightly relaxed and allowed minimum AoA -35° (instead of default -30°) and maximum AoA +35° (instead of default +35°)

Random flux uncertainty: Finkelstein and Sims (2001, Definition of the ITS: Cross-correlation first corssing 1/e, Maximum correlation period: 10s)

Spectral Analysis and Corrections#

Spectra and (co)spectra calculation: Binned (co)spectra not available, Use power-of-two samples to speed up the FFT: YES, Lowest noise frequency: all gases 1Hz
Spectra and Cospectra QA/QC: in the following the three settings for “Minimum, unstable”/”Minimum, stable”/”Maximum”, respectively. In particular, note the settings for CH₄ and N₂O. Settings for these two had to be strongly adjusted to accomodate the often very low fluxes at the site.
- Friction velocity 0.2/0.05/5 m s^-1
- Sensible heat flux 20/5/1000 W m^-2
- Latent heat flux 20/3/1000 W m^-2
- CH₄ 0.01/0.005/20 µmol m² s^-1
- N₂O (4^th gas) 0.01/0.005/20 µmol m² s^-1
Filter (co)spectra according to Vickers and Mahrt (1997) test results: NO
Filter (co)spectra according to micrometeorological quality test results (Mauder and Foken, 2004): Low data quality (flag value = 2): YES, Moderate data quality (flag = 1): NO
Spectral Correction Options:
- Low frequency range: Analytic correction of high-pass filtering effects (Moncrieff at al. 2004): YES
- High frequency range: Correction of low-pass filtering effects: YES, for LI-7500 fluxes Horst (1997), for QCL or LGR Fratini et al. (2012) because these are closed-path instruments
- Spectral assessment file (SAF) was not available for most time periods. Generally calculated for each year, or for each distinct time period that required separate processing (e.g. in case of setup changes). In case of very short time periods the SAF from neighboring time periods was used. For example: when flux calculations were done for a short time period in May 2020 (few days), the SAF from the following time period was used (SAF based on 6+ months).
- Assessment of high-frequncy attenuation: all gases: Lowest frequency 0.005 Hz, Highest frequency 2 Hz
- We also output full w/Ts cospectra for final flux calculations in Level-1

Output files#

Spectra and cospectra output: Reduced spectra and ogives YES all, Full length spectra: Ts YES for Fratini at al. (2012) correction, Full length cospectra: NO all
Processed raw data: all statistics, no time series

Used Level-1 files#

List of EddyPro output file names. Results from these files were used to create the dataset.

IRGA#

Notebook

31.0_FLUXES_L1_IRGA Notebook that was used to merge all IRGA Level-1 result files.

2005_1_IRGA_eddypro_CH-CHA_FR-20240730-112428_fluxnet_2024-07-30T182651_adv.csv
2006_1_IRGA_eddypro_CH-CHA_FR-20240730-112420_fluxnet_2024-07-31T080813_adv.csv
2007_1_IRGA_eddypro_CH-CHA_FR-20240730-112410_fluxnet_2024-07-31T064739_adv.csv
2008_1_IRGA_eddypro_CH-CHA_FR-20240730-112401_fluxnet_2024-07-31T085214_adv.csv
2009_1+3_eddypro_CH-CHA_FR-20240919-144235_fluxnet_2024-09-20T054007_adv.csv
2009_2_eddypro_CH-CHA_FR-20240919-144503_fluxnet_2024-09-19T181458_adv.csv
2010_1+3_IRGA_eddypro_CH-CHA_FR-20240728-190324_fluxnet_2024-07-29T140100_adv.csv
2010_2_IRGA_eddypro_CH-CHA_FR-20240730-112352_fluxnet_2024-07-30T160648_adv.csv
2011_1_IRGA_eddypro_CH-CHA_FR-20240728-190344_fluxnet_2024-07-29T153514_adv.csv
2012_1_IRGA_eddypro_CH-CHA_FR-20240730-112342_fluxnet_2024-07-30T134628_adv.csv
2012_2_IRGA_eddypro_CH-CHA_FR-20240728-190409_fluxnet_2024-07-29T165401_adv.csv
2013_1_IRGA_eddypro_CH-CHA_FR-20240727-210036_fluxnet_2024-07-29T001913_adv.csv
2014_1_IRGA_eddypro_CH-CHA_FR-20240727-210133_fluxnet_2024-07-29T001656_adv.csv
2015_1_IRGA_eddypro_CH-CHA_FR-20240727-210058_fluxnet_2024-07-28T091257_adv.csv
2015_2_IRGA_eddypro_CH-CHA_FR-20240727-210122_fluxnet_2024-07-28T151141_adv.csv
2016_1+3_IRGA_eddypro_CH-CHA_FR-20240730-112331_fluxnet_2024-07-31T065812_adv.csv
2016_2_IRGA_eddypro_CH-CHA_FR-20240730-112320_fluxnet_2024-07-30T165223_adv.csv
2017_1+2+3_IRGA_eddypro_CH-CHA_FR-20240727-210107_fluxnet_2024-07-29T012549_adv.csv
2018_1+2_IRGA_eddypro_CH-CHA_FR-20240727-210213_fluxnet_2024-07-28T230450_adv.csv
2018_3_IRGA_eddypro_CH-CHA_FR-20240730-112310_fluxnet_2024-07-30T132817_adv.csv
2018_4_IRGA_eddypro_CH-CHA_FR-20240730-112301_fluxnet_2024-07-30T144534_adv.csv
2019_1+2+3+4+5_IRGA_eddypro_CH-CHA_FR-20240727-210200_fluxnet_2024-07-29T014548_adv.csv
2020_1+2+3_IRGA_eddypro_CH-CHA_FR-20240727-210028_fluxnet_2024-07-28T094044_adv.csv
2020_4_IRGA_eddypro_CH-CHA_FR-20240730-112251_fluxnet_2024-07-30T121132_adv.csv
2020_5_IRGA_eddypro_CH-CHA_FR-20240727-210053_fluxnet_2024-07-28T161420_adv.csv
2021_1_IRGA_eddypro_CH-CHA_FR-20240727-005940_fluxnet_2024-07-27T135432_adv.csv
2021_2_IRGA_eddypro_CH-CHA_FR-20240727-010348_fluxnet_2024-07-27T120109_adv.csv
2022_1+2_IRGA_eddypro_CH-CHA_FR-20240726-181747_fluxnet_2024-07-27T080504_adv.csv
2022_3_IRGA_eddypro_CH-CHA_FR-20240726-181749_fluxnet_2024-07-27T051830_adv.csv
2023_01+eddypro_CH-CHA_FR-20250303-102224_fluxnet_2025-03-04T015312_adv.csv
2023_02+eddypro_CH-CHA_FR-20250304-093959_fluxnet_2025-03-04T110501_adv.csv
2024_IRGA_eddypro_CH-CHA_FR-20250124-134851_fluxnet_2025-01-25T080153_adv.csv

QCL/LGR#