-
Notifications
You must be signed in to change notification settings - Fork 5
Level 6
Level 6, or L6 for short, is the stage where PyFluxPro partitions the net ecosystem exchange (NEE) into ecosystem respiration (ER) and gross primary productivity (GPP). At L6, PyFluxPro also calculates evapo-transpiration (ET) from the latent heat flux and prepares summaries of the data at daily, monthly and annual time steps. The summaries are output as both netCDF and Excel files.
As with other processing levels, options for the partitioning process are specified in a control file that can be edited in the PyFluxPro GUI. Templates for L6 control files are in the PyFluxPro/controlfiles/templates/L6 folder.
The method used by PyFluxPro to partition NEE is as follows:
- At L3, PyFluxPro will, if requested, add the storage term to Fco2, the turbulent flux of CO2. The storage term can be;
- Estimated from profile measurements of CO2 concentration outside PyFluxPro and read in at L1 or;
- Estimated by PyFluxPro from the time derivative of the CO2 concentration at the height of the IRGA.
- The storage corrected Fco2 is filtered for low turbulence conditions prior to gap filling at L5.
- At L6, observed ER is taken to be the nocturnal, u* filtered (done at L5), storage corrected (done at L3 if requested) CO2 flux, Fco2.
- A respiration model is trained on this observed ER. PyFluxPro offers 3 methods for modelling respiration:
- The SOLO neural network. This method can use an arbitrary set of drivers to predict ER.
- The Lloyd-Taylor method (see Lloyd and Taylor, 199?). This is similar to the night time method from FluxNet 2015 and uses the air temperature Ta as the driver.
- The Lasslop method (see Lasslop et al, 2010). This is similar to the day time method from FluxNet 2015 and uses the air temperature Ta as the driver.
- The respiration model trained on observed ER is then applied to the gap-filled time series of the driver(s) to predict ER for all time steps in the data set. The final ER time series is constructed from the observations and the model predictions.
- The final NEE time series is constructed from the observations, NEE values from the L5 gap filling during the day and predictions from the L6 ER model during the night.
- GPP is then calculated from the NEE and ER time series as; GPP = -1*NEE + ER
Two other quantities are calculated at L6:
- Net ecosystem productivity (NEP) as; NEP = -1*NEE
- Evapo-transpiration (ET) as; ET = Fe/Lv where Lv is the latent heat of vapourisation.
Finally, summaries of the gap filled and partitioned data are calculated at L6 as daily, monthly and annual averages or totals and as cumulative totals over individual years and the whole data set. The summary data is output in both netCDF files and Excel workbooks.