Projected land photosynthesis constrained by changes in the seasonal cycle of atmospheric CO2#

Overview#

Selected figures from Wenzel et al. (2016) are reproduced with recipe_wenzel16nat.yml. Gross primary productivity (gpp) and atmospheric CO2 concentrations at the surface (co2s) are analyzed for the carbon cycle - concentration feedback in the historical (esmHistorical) and uncoupled (esmFixCLim1, here the carbon cycle is uncoupled to the climate response) simulations. The recipe includes a set of routines to diagnose the long-term carbon cycle - concentration feedback parameter (beta) from an ensemble of CMIP5 models and the observable change in the CO2 seasonal cycle amplitude due to rising atmospheric CO2 levels. As a key figure of this recipe, the diagnosed values from the models beta vs. the change in CO2 amplitude are compared in a scatter plot constituting an emergent constraint.

Available recipe and diagnostics#

Recipes are stored in recipes/

  • recipe_wenzel16nat.yml

Diagnostics are stored in diag_scripts/carbon_ec/

  • carbon_beta: (1) scatter plot of annual gpp vs. annual CO2 and (2) barchart of gpp(2xCO2)/gpp(1xCO2); calculates beta for emergent constraint (carbon_co2_cycle.ncl)

  • carbon_co2_cycle.ncl: (1) scatter plot of CO2 amplitude vs. annual CO2, (2) barchart of sensitivity of CO2 amplitude to CO2, (3) emergent constraint: gpp(2xCO2)/gpp(1xCO2) vs. sensitivity of CO2 amplitude to CO2, (4) probability density function of constrained and unconstrained sensitivity of CO2 amplitude to CO2

User settings#

Note

Make sure to run this recipe setting output_file_type: pdf in the config_user.yml file or using the CLI flag --output_file_type=pdf.

  1. Script carbon_beta.ncl

    Required Settings (scripts)

    • styleset: project style for lines, colors and symbols

    Optional Settings (scripts)

    • bc_xmax_year: end year to calculate beta (default: use last available year of all models)

    • bc_xmin_year: start year to calculate beta (default: use first available year of all models)

    Required settings (variables)

    none

    Optional settings (variables)

    none

  2. Script carbon_co2_cycle.ncl

    Required Settings (scripts)

    • nc_infile: path of netCDF file containing beta (output from carbon_beta.ncl)

    • styleset: project style for lines, colors and symbols

    Optional Settings (scripts)

    • bc_xmax_year: end year (default = last year of all model datasets available)

    • bc_xmin_year: start year (default = first year of all model datasets available)

    Required settings (variables)

    • reference_dataset: name of reference datatset (observations)

    Optional settings (variables)

    none

Variables#

  • co2s (atmos, monthly mean, plev longitude latitude time)

  • gpp (land, monthly mean, longitude latitude time)

Observations and reformat scripts#

  • ESRL: Earth System Research Laboratory, ground-based CO2 measurements

References#

  • Wenzel, S., Cox, P., Eyring, V. et al., 2016, Projected land photosynthesis constrained by changes in the seasonal cycle of atmospheric CO2. Nature 538, 499501, doi: doi.org/10.1038/nature19772

Example plots#

../_images/fig_1.png

Fig. 126 Comparison of CO2 seasonal amplitudes for CMIP5 historical simulations and observations showing annual mean atmospheric CO2 versus the amplitudes of the CO2 seasonal cycle at Pt. Barrow, Alaska (produced with carbon_co2_cycle.ncl, similar to Fig. 1a from Wenzel et al. (2016)).#

../_images/fig_2.png

Fig. 127 Barchart showing the gradient of the linear correlations for the comparison of CO2 seasonal amplitudes for CMIP5 historical for at Pt. Barrow, Alaska (produced with carbon_co2_cycle.ncl, similar to Fig. 1b from Wenzel et al. (2016)).#

../_images/fig_3.png

Fig. 128 Emergent constraint on the relative increase of large-scale GPP for a doubling of CO2, showing the correlations between the sensitivity of the CO2 amplitude to annual mean CO2 increases at Pt. Barrow (x-axis) and the high-latitude (60N - 90N) CO2 fertilization on GPP at 2xCO2. The red line shows the linear best fit of the regression together with the prediction error (orange shading), the gray shading shows the observed range (produced with carbon_co2_cycle.ncl, similar to Fig. 3a from Wenzel et al. (2016)).#