Level-0: Preliminary flux calculations

Level-0: Preliminary flux calculations#

Info

Results from preliminary flux calculations with relaxed processing settings assist in finding the most appropriate settings for final flux calculations (Level-1).

Level-0 flux results help to find the best settings for the final flux calculations. Generally, these preliminary results are used:

  • to check whether the flux processing works,

  • to check the wind direction across years,

  • to determine appropriate time windows for lag search for final flux calculations.

More details can be found in the documentation of the Flux Processing Chain

OPENLAG runs to determine final lag ranges#

Notebook

02_L0_timelags_check

  • The lag between turbulent departures of wind and the gas of interest was first determined in a relatively wide time window (called OPENLAG):

    • IRGA OPENLAG time window: between -1s and +10s. For IRGA, the goal was to find an appropriate nominal time lag, and to determine a time window for lag search as narrow as possible.

    • QCL OPENLAG time window: between -1s and +10s, and a second OPENLAG run between -1s and +5s. The second OPENLAG run was done because the first run showed that the time lags accumulated in a narrower range below +5s. For QCL fluxes, constant time lags were used for the final flux calcs, so here in the OPENLAG runs it was very important to get the value for the constant lag as correct as possible. The final lag range was then set according to these narrower results.

  • The value -1s was used as the starting value for the time window because EddyPro had issues with a start value of 0s, in that case the lag search started e.g. at +1s for some reason (maybe bug)

  • The results from the OPENLAG runs were used to define a time window as narrow as possible (IRGA) for the final flux calculations (Level-1)

    • The histograms of found OPENLAG time lags were inspected to determine whether there was a the histogram bin with peak distribution

    • IRGA: peak of distribution was used as nominal (default) lag, time ranges around this peak were used as time window (table)

    • QCL: peak of distrubution was used as constant lag for the respective time period, no time window used, all used lags were constant (table)

Fig. 2 Results for the OPENLAG run for CO2 (Level-0 fluxes) in 2024. The time lag between turbulent departures of vertical wind and CO2 molar densities of the open-path IRGA was first searched in a time window between -1s and 10s, then the window was narrowed down to get a clearer histogram peak for the most frequently found time lag. Level-0 flux calculations use no nominal (default) time lag, only simple covariance maximization.These results were used to define a narrow time window for final flux calculations in Level-1. In this example, for Level-1 flux calculations, the time lag with peak distribution was used as the nominal (default) time lag (0.30s), the time window for lag search was set to between +0.05s and +0.5s.#

Fig. 3 Histogram and time series of found time lags between turbulent departures of vertical wind and N2O mixing ratios. Results from the OPENLAG run (Level-0 fluxes) in 2021. The histogram shows a clear bimodal distribution, which indicates that the time lag between the two scalars is not constant. In this example, there are two distinct time periods: the first time period has shorter lags and ranges from January to 22 July; the second time period has longer lags and ranges from 23 July to 31 December. In this example, the fluxes for the two time periods are calculated separately, with their own respective constant time lags. Time lags get noisier towards the end of the year due to (very) low N2O fluxes and therefore low signal-to-noise ratios. However, lags after 31 December 2021, during the following year, are still in the same range, therefore our best estimate is that the lags remain essentially the same during the noisy period.#

Table 4 IRGA nominal (default) time lags and size of the lag search windows for different time periods in seconds. Used for CO2 and H2O (LE) in final flux calculations.#

time period

CO2 IRGA

H2O IRGA

Notes

2005_X

0.20, 0.05-0.40

same as CO2

2006_X

0.25, 0.05-0.40

same as CO2

2007_X

0.25, 0.05-0.40

same as CO2

2008_X

0.25, 0.05-0.40

same as CO2

2009_X

0.25, 0.05-0.40

same as CO2

2010_X

0.30, 0.05-0.50

same as CO2

2011_X

0.30, 0.05-0.55

same as CO2

2012_X

0.35, 0.05-0.55

same as CO2

first time period QCL

2013_X

0.30, 0.05-0.55

same as CO2

2014_X

0.30, 0.05-0.55

same as CO2

2015_X

0.30, 0.05-0.55

same as CO2

2016_X

0.35, 0.05-0.55

same as CO2

2017_X

0.30, 0.05-0.55

same as CO2

2018_1+2+3+4

0.25, 0.05-0.50

same as CO2

2019_1+2+3+4+5

0.30, 0.05-0.45

same as CO2

2020_1+2+3+4+5

0.30, 0.05-0.45

same as CO2

2021_1+2

0.25, 0.05-0.45

same as CO2

2022_1+2+3

0.25, 0.05-0.45

same as CO2

2023_1

0.30, 0.05-0.50

same as CO2

2024_1

0.30, 0.05-0.50

same as CO2

Table 5 QCL and LGR constant time lags (seconds) for N2O and CH4 used in final flux calculations. In addition, the range where most time lags were found is given.#

time period

N2O QCL LGR

CH4 QCL LGR

H2O QCL LGR

Notes

2012_X

0.85, 0.70-1.50

0.85, 0.70-1.50

1.30, 1.00-3.00

first time period QCL

2013_X

1.15, 0.85-1.70

1.10, 0.85-1.80

2.10, 1.30-3.30

2014_X

1.25, 0.95-1.60

1.15, 0.90-1.80

2.00, 1.30-3.30

2015_X

1.35, 0.70-2.00

1.25, 0.70-2.00

1.95, 1.20-4.00

2016_2

0.85, 0.65-2.00

1.00, 0.60-2.00

2.45, 1.80-4.50

2016_1+3_2017_1+2

1.15, 0.70-1.95

0.95, 0.70-2.00

1.95, 1.50-3.30

2017_1+2: no H2O lag visible

2017_3_2018_1

1.60, 1.20-2.35

1.65, 1.10-2.45

2.50, 1.50-4.00

only few data, no H2O lag visible

2018_2

n.a.

n.a.

n.a.

2018_2: no data for QCL

2018_3

2.00, 1.25-2.55

2.05, 1.25-2.55

4.45, 4.00-6.00

no H2O lag visible

2018_4_2019_1+3

1.45, 1.00-2.50

1.25, 1.00-2.50

n.a.

no H2O

2019_2

6.30, 4.00-10.00

6.80, 4.00-10.00

n.a.

not clear in Mar and Apr; no H2O

2019_4

1.55, 0.90-2.35

1.35, 0.90-2.20

n.a.

no H2O

2019_5

1.35, 1.00-2.50

1.30, 1.00-2.50

n.a.

no H2O

2020_1+2

1.45, 1.00-2.50

1.55, 1.00-2.50

n.a.

no H2O

2020_3

0.70, 0.40-0.90

0.65, 0.45-0.90

n.a.

no H2O

2020_4+5_2021_1

0.60, 0.40-0.90

0.65, 0.45-0.90

0.70, 0.60-2.00

last time period QCL, H2O available again since 2020_4

2021_2_2022_1

1.75, 1.50-3.30

1.75, 1.50-3.30

1.80, 1.65-6.00

LGR, lag fluctuates within these ranges

Check wind direction across years#

Notebook

03_L0_winddir_check

I compared histograms of wind directions between 2005 and 2024 using Level-0 fluxes and found that a sonic orientation of 7° offset to north yields very similar results across years. It is therefore possible the the sonic orientation across all years was always close to 7°.

Here are results from a comparison of annual wind direction histograms (with bin width of 1°) to a reference period (2006-2009), all wind directions were calculated with a north offset of 7°, then a histogram was calculated for each year. The OFFSET describes how many degrees have to be added (or subtracted) to the half-hourly wind direction to yield a histogram that is most similar to the reference. All OFFSETS are small, which indicates that the wind directions are in good agreement (table).

Table 6 Wind direction offsets (in degrees) compared to a reference period (2006-2009) from Level-0 OPENLAG runs.#

YEAR

OFFSET (°)

2005

1.0

2006

0.0

2007

-2.0

2008

-2.0

2009

0.0

2010

2.0

2011

6.0

2012

1.0

2013

1.0

2014

1.0

2015

1.0

2016

3.0

2017

4.0

2018

1.0

2019

-1.0

2020

-1.0

2021

-1.0

2022

1.0

2023

-2.0

2024

1.0
Contents