Fig 1.
Change in atmospheric CO2 concentration, and thus surface warming, differs significantly across the different pathways for the same amount of emitted carbon.
a The colored lines depict the four different pathways with constant (blue), negative parabolic (green), linearly decreasing (yellow), and exponentially increasing carbon emissions (red). The pathways are either realized over a time period of 100 or 200 years (x-axis). The annual emission rates are given for the 200 year period as marked by the * (y-axis). b The colored lines depict the time integrals of the four pathways shown in a, where all accumulate to 1.2 Eg emitted carbon in the final year. c Relationship between cumulative carbon emissions (x-axis) and the change in atmospheric CO2 concentration (Δ[CO2]) is shown for 200 years, where the colors refer to the four different pathways analogous to a. Each circle refers to one year and the arrows annotate the final Δ[CO2] in each model experiment. See S1 Fig for a comparison of 100 vs 200 year runs. d Transient Climate Response to Cumulative Emissions (TCRE) for different pathways (x-axis) and simulation periods (colors). TCRE is estimated using the conventionally used linear regression method [13]. Shaded dots exhibit the spread in the estimates of the final five years when 1.2 Eg C have been emitted as well as among different realizations. Pointplot with whiskers show the mean and standard deviation of the spread. Companion S2 Fig shows Land TCRE.
Fig 2.
Change in global mean surface air temperature across the different pathways and different periods for the same amount of emitted carbon.
a–d Relationship between cumulative carbon emissions (x-axis) and the change in global mean air temperature (ΔTair) is shown, where the colors refer to the four different pathways analogous to Fig 1 and the markers refer to the different time periods. To reduce inter-annual variability, five-year averages across the three realizations are displayed. The arrows annotate the final ΔTair in each model experiment. The dotted black line facilitates comparison between the pathways.
Table 1.
Change in annual mean surface air temperature at the end of the emission phase compared to pre-industrial conditions, aggregated globally and separated for land and ocean.
Fig 3.
Decomposition of TCRE into planetary heat uptake, carbon cycling, and physical feedback parameter.
a The evolution of TCRE in the four different emission pathways over the last five decades in the 200-year simulations. Each thick line represents the ensemble average of three realizations per experiment smoothed by a low-pass filter (Savitzky-Golay filter). Following the black arrows, TCRE is first decomposed into the airborne fraction of emissions (expressed in cumulative terms, i. e., atmospheric carbon change over cumulative emissions; b) and the Transient Climate Response (TCR, c). TCR is further decomposed into the radiative forcing of atmospheric carbon (d) and the thermal response (e), which is again decomposed into the fraction of radiative forcing warming the surface (f) and the inverse climate feedback parameter (g). The equations displayed at the top of a, c and e describe the three steps of the TCRE factorization and are explained in more detail in the Eqs 5 to 9 in Materials and Methods.
Fig 4.
Vegetation dynamics explain the pathway dependence in the terrestrial carbon cycle.
a Global map of the difference in the fraction of forested area between the L and E pathways (yellow and red in b and c, respectively) at the end of the emission phase, calculated with MPI-ESM1–2-LR in the 200-year runs. b Change in vegetated area for forest, grass- and shrublands, and total vegetation for the different emission pathways. The whiskers represent the uncertainty between the different realizations. c Change of forested area across the four pathways as a function of time for both the emission and relaxation phases. Companion S3 Fig shows changes in associated carbon pools.
Fig 5.
Evolution of atmospheric excess carbon after emission stop and land carbon sensitivity to different CO2 growth rates in CMIP ensembles.
a Different colored dots depict different CMIP6 models in ZecMIP, where the lines refer to the best fit using a double exponential decay function (see Materials and methods). b as in a but the colors refer to the different emission pathways and the ZecMIP run of the MPI-ESM1–2-LR model. c Change in the land carbon pool at latitudes above 60°N per a 100 ppm change in atmospheric CO2 across three different future scenarios in CMIP6 reflecting different CO2 growth rates (x-axis; see Materials and methods). The black dot refers to the multi-model mean value for each growth rate, where the whiskers represent the standard deviation. d as in c but for CMIP5. Companion S5 Fig shows changes in green leaf area index in CMIP6 and CMIP5 simulations.
Fig 6.
State of the carbon system, sink efficiency, and TCRE across the pathways.
a Cumulative land and ocean sinks are juxtaposed across all pathways and both simulation periods. The markers denote the final state at the end of the emission phase of 1.2 Eg C. The dashed line indicates equal magnitude of cumulative land and ocean sinks. The black line refers to the average relationship inferred from GCP estimates, where the gray dots follow plus/minus the joint uncertainty for both cumulative land and ocean sink. b As in a, but for the relaxation phase. c Sink efficiency is plotted as function of the cumulative carbon sink for both emission and relaxation phase. The lines represent a nonparametric LOWESS fit, with the solid (dashed) line referring to the 200-(100-)year simulations in the different pathways indicated by different colors. The black lines indicate the relationship when all pathways are fitted at once. d TCRE is plotted as function of the cumulative carbon sink, i. e., the state of the carbon system. The colored lines represent a linear fit for the four different pathways. The black lines indicate the relationship when all pathways and both periods are fitted linearly at once. The grey shading reflects uncertainty based on an 95% confidence interval estimated using bootstrapping. Companion S7 Fig shows the land-ocean sink relationship during the emission and relaxation phase for the EMIC CLIMBER-2.