Offsetting unabated agricultural emissions with CO2 removal to achieve ambitious climate targets

To calculate the required CDR rates, we use two different approaches. The first approach (“GWP100”), depicted in Eq 1, sets CDR rates ECDR in (GtCO2/yr) equal to the CO2-equivalent of additional agricultural emissions of species i at every point in time, using a GWP100 value for methane of 28 [3]. The additional agricultural emissions are calculated from the difference in emissions between the SSP1-2.6 and the alternative scenarios ðEscenario;iagri ESSP1 2:6;iagri Þ with i being either methane or nitrous oxide emissions, in GtCH4/yr or GtN2O/yr): ECDR ¼ PGWP100i � ðEscenario;iagri ESSP1 2:6;iagri Þ ð1Þ As the use of GWP100 for short-lived greenhouse gases has limitations [47–51], we explore the use of two additional alternative metrics: the GWP20 and the GWP�. We do not explore the use of GTP100 since the metric leads to larger error, when substituting methane with CO2 emissions, than the GWP100 [56]. Since the use of the GWP to calculate equivalent CO2 emis- sions reductions is accurate only for short time horizons [48, 53], we test the effects of using the GWP metric over a time frame of 20 years (GWP20). GWP� denotes an alternative usage of the GWP relating cumulative CO2 emissions to date with a change in the rate of emissions of short-lived climate pollutants [55]. We calculate GWP20-based CDR rates as per Eq 1, using GWP20 instead of GWP100. Differently, we calculate GWP�-based CDR as per Eq 2 [55]. ECDR ¼ DðEscenario;CH4agri ESSP1 2:6;CH4agri Þ Dt � GWP100 � 100 ð2Þ In a second offsetting approach, we abandon conversion metrics altogether, and instead cal-

culate CDR rates through explicit modelling of effective radiative forcing (we call this the “ERF approach”). In the ERF approach, we explicitly compute CDR requirements using the Finite Amplitude Impulse Response (FaIR) model [64, 65], an open-source reduced complexity car- bon-cycle, atmospheric composition and climate model [66]. Starting from input climate-forc- ing species concentrations or emissions, FaIR calculates the corresponding ERF and temperatures. We numerically invert FaIR to calculate the CDR rates needed to maintain the Fig 1. Methane (panel a, “CH4”) and nitrous oxide (panel b, “N2O”) emission pathways under the SSP1-2.6 and alternative scenarios (constant and worst-case emissions). Agricultural emissions differ between the three scenarios (SSP1-2.6, constant, and worst-case) while non-agricultural emissions are the same in all three scenarios (and corresponding to SSP1-2.6 emissions). The data used to reproduce this figure are publicly available in the IIASA SSP database [60–63]. https://doi.org/10.1371/journal.pone.0247887.g001 PLOS ONE Offsetting future agricultural emissions with CO2 removal PLOS ONE | https://doi.org/10.1371/journal.pone.0247887 March 17, 2021 5 / 19