EGU General Assembly 2023
© Author(s) 2023. This work is distributed under
the Creative Commons Attribution 4.0 License.

Synthetic data experiment to test the accuracy of methods estimating carbon uptake period from atmospheric CO2 time-series 

Theertha Kariyathan1,2, Ana Bastos1, Markus Reichstein1, Wouter Peters2, and Julia Marshall3
Theertha Kariyathan et al.
  • 1Max Planck Institute for Biogeochemistry, Jena, Germany
  • 2Wageningen University and Research, Wageningen, Netherlands
  • 3Deutsches Zentrum für Luft- und Raumfahrt (DLR), Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany

Atmospheric CO2 measurements from background sites across the Northern Hemisphere have been used to study the changes in the carbon uptake period (CUP) i.e., when plants are able to grow and assimilate carbon from the atmosphere. Previous studies that use CO2 dry air mole fraction data diagnosed CUP using zero-crossing dates (ZCD, when the detrended seasonal cycle switches from positive to negative sign and vice versa). The CUP can also be estimated using the first derivative of the CO2 seasonal cycle. In previous work we show that applying the first derivative method to an ensemble of fitted CO2 mole fraction curves provides better constraints to the CUP by considering year-to-year uncertainty in estimates across the ensemble members. We call this the ensemble of first derivative method (EFD method).  In addition to curve fitting uncertainty and year-to-year flux variability, atmospheric transport might explain a significant portion of observed CO2 variations at various surface stations, affecting the interpretation of the CUP and similar metrics.

Hence, in this study we examine how atmospheric transport of fluxes, and spatial variations in the start and ending dates of carbon uptake, smooth the signal in atmospheric CO2 and affect the CUP estimates when using remote background observation sites to interpret actual fluxes. We use a synthetic data experiment where idealized NEE fluxes are transported forward (with atmospheric transport model TM3 (Heimann and Körner, 2003) and fixed year meteorology) and the atmospheric concentrations are sampled at the location of the measurement sites. A fixed year from the Jena CarboScope Inversion (Rödenbeck et al., 2003, doi:10.17871/CarboScope-sEXTocNEET_v2022) was used to generate an idealized NEE flux time series with no interannual variability in the CUP at any given pixel. Then, we prescribe changes in the CUP of NEE flux to Northern Hemisphere land pixels with clear seasonal cycles and evaluate the accuracy of the ZCD and EFD methods in capturing this known change from CUP in the surface fluxes, from the resulting CO2 mixing ratio obtained from the forward transport run.

We find that CUP changes estimated by both EFD and ZCD based on CO2 measurements are smaller by a factor of 2-4 than the perturbations applied in NEE space, and that the EFD method is more sensitive to surface CUP changes than the ZCD. This " dampening" factor varies across sites, depending on the mixing of spatially varying NEE signals with differing CUP timing which integrate to a reduced atmospheric expression of CUP. We further analyse the contribution of 1) atmospheric transport by comparing simulation that uses inter annually varying meteorology 2) different TransCom-3 regions to CUP variations by selectively manipulating NEE flux from a region and repeating the experiment.


Heimann, H. and Körner, S. (2003). The global atmospheric tracer model tm3. Technical Reports- Max-Planck-Institut f ̈ur Biogeochemie 5, 5:131.

Rödenbeck, C., Houweling, S., Gloor, M., and Heimann, M. (2003). Co2 flux history 1982–2001 inferred from atmospheric data using a global inversion of atmospheric transport. Atmospheric Chemistry and Physics, 3(6):1919–1964. doi: 10.17871/CarboScope-sEXTocNEET_v2022.

How to cite: Kariyathan, T., Bastos, A., Reichstein, M., Peters, W., and Marshall, J.: Synthetic data experiment to test the accuracy of methods estimating carbon uptake period from atmospheric CO2 time-series , EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-7261,, 2023.