Response of atmospheric pCO 2 to a strong AMOC weakening under low and high emission scenarios

Abstract

The Earth System is warming due to anthropogenic greenhouse gas emissions which increases the risk of passing a tipping point in the Earth System, such as a collapse of the Atlantic Meridional Overturning Circulation (AMOC). An AMOC weakening can have large climate impacts which influences the marine and terrestrial carbon cycle and hence atmospheric pCO $_2^{}$ . However, the sign and mechanism of this response are subject to uncertainty. Here, we use a state-of-the-art Earth System Model, the Community Earth System Model v2 (CESM2), to study the atmospheric pCO $_2^{}$ response to an AMOC weakening under low (SSP1-2.6) and high (SSP5-8.5) emission scenarios over the years 2015-2100. A freshwater flux anomaly in the North Atlantic strongly weakens the AMOC, and we simulate a weak positive pCO $_2^{}$ response of 0.45 and 1.3 ppm increase per AMOC decrease in Sv for SSP1-2.6 and SSP5-8.5, respectively. For SSP1-2.6 this response is driven by both the oceanic and terrestrial carbon cycles, whereas in SSP5-8.5 it is solely the ocean that drives the response. However, the spatial patterns of both the climate and carbon cycle response are similar in both emission scenarios over the course of the simulation period (2015-2100), showing that the response pattern is not dependent on cumulative CO $_2^{}$ emissions up to 2100. Though the global atmospheric pCO $_2^{}$ response might be small, locally large changes in both the carbon cycle and the climate system occur due to the AMOC weakening, which can have large detrimental effects on ecosystems and society.

Supplementary information: The online version contains supplementary material available at 10.1007/s00382-024-07295-y.

Keywords: AMOC weakening; Atlantic meridional overturning circulation; Atmospheric pCO 2 ; Carbon cycle; Climate change; Marine biosphere.

Fig. 1

a AMOC strength at 26.5$_{}^{\circ }$N in Sv, b GMST in $_{}^{\circ }$C, and c atmospheric CO$_2^{}$ concentration in ppm for the period 2020–2100 smoothed with a 5 year moving mean. In (a–c) blue lines represent the control (CTL) simulations, and orange lines the HOS simulations. (d–f) as in (a–c) but for the difference between the HOS simulations and the control simulations. In all subplots dashed lines represent SSP1-2.6 (126) and solid lines SSP5-8.5 (585)

Fig. 2

Results for Surface Air Temperature (SAT) in $_{}^{\circ }$C. The top row (a–c) is for SSP1-2.6, and the bottom row (d–f) for SSP5-8.5. The left column (a, d) represents the average over 2016–2020 in the control simulations. The middle row (b, e) represents the difference between the average of 2096–2100 and 2016–2020 for the control simulations. The right row (c, f) represents the difference between the HOS and CTL simulations averaged over 2096–2100. Note the different scaling between b and e

Fig. 3

Results for the oceanic CO$_2^{}$ uptake integrated over the entire simulation period in kg C m$_{}^{-2}$. The top row (a–c) represents SSP1-2.6 and the bottom row (d–f) represents SSP5-8.5. The left column (a, d) represents the uptake in the control simulations, the middle column (b, e) the uptake in the HOS simulations, and the right column (c, f) the difference between the HOS and CTL simulations. In a, b, d, and e positive values (brown colors) represent net uptake, and negative values (blue colors) represent net outgassing

Fig. 4

a Cumulative uptake of CO$_2^{}$ in the ocean from 2020 onward in PgC. b The difference in the cumulative oceanic CO$_2^{}$ uptake between the HOS and CTL simulations. c Difference in the cumulative oceanic CO$_2^{}$ uptake between the HOS and CTL simulations in SSP1-2.6 for different ocean basins. d As in (c) but for SSP5-8.5. In (a) blue lines represent the CTL simulations, and the orange lines the HOS simulations. In all subplots dashed lines represent SSP1-2.6 and solid lines SSP5-8.5. Negative values in b–d represent reduced uptake in the HOS simulations compared to the CTL simulations. Results are smoothed with a 5 year moving mean

Fig. 5

Response of specific variables to the AMOC weakening, i.e. the HOS minus the CTL simulations, zoomed in on the North Atlantic and Arctic Ocean. Columns 1 and 3 represent SSP1-2.6 and columns 2 and 4 SSP5-8.5. The black lines represent the 0.15 sea ice fraction averaged over 2096–2100 for the CTL simulation (dashed lines) and HOS simulation (solid lines). Panels a and b represent integrated ocean uptake over the entire simulation period, the other panels represent averages over 2096–2100. The variables shown here are the gas exchange (c, d), surface pH (e, f), SSTs (g, h), SSSs (i, j), DIC in the top 150 m (k, l), Alk in the top 150 m (m, n), NPP (o, p), and maximum mixed layer depth (q, r)

Fig. 6

Results for the CO$_2^{}$ exchange with the land integrated over the entire simulation period in kg C m$_{}^{-2}$. The top row (a–c) represents SSP1-2.6 and the bottom row (d–f) represents SSP5-8.5. The left column (a, d) represents the uptake in the control simulations, the middle column (b, e) the uptake in the HOS simulations, and the right column (c, f) the difference between the HOS and CTL simulations. In a, b, d, and e green colors represent net CO$_2^{}$ uptake by the land, and red colors represent net emissions into the atmosphere

Fig. 7

a Cumulative uptake of CO$_2^{}$ on the land from 2020 onward in PgC. b Difference in the cumulative land uptake of CO$_2^{}$ between the HOS and CTL simulations. Blue lines represent the control simulations, and the orange lines the HOS simulations. In both subplots dashed lines represent SSP1-2.6 and solid lines SSP5-8.5. Negative values in b represent reduced uptake in the HOS simulations compared to the CTL simulations. Results are smoothed with a 5 year moving mean

Fig. 8

Summarizing figure with dominant mechanisms included for SSP1-2.6 (a) and SSP5-8.5 (b). a and b represent results from HOS minus the CTL simulations. The sea-ice edge is taken as where the ice fraction is 0.25 and denoted by the purple lines, where dashed lines represent the CTL simulations and solid lines the HOS simulations. The bar at the left shows the difference in zonal mean surface air temperature averaged over 2096–2100 between HOS and CTL. The scaling of this bar is between $-$2.5 $_{}^{\circ }$C (dark blue) and 2.5 $_{}^{\circ }$C (dark red). c The difference between SSP5-8.5 (b) and SSP1-2.6 (a) for the regions where (b) is negative. Negative values represent a higher negative anomaly in SSP5-8.5 compared to SSP1-2.6. (d) as in (c) but for positive anomalies. Positive values represent a higher positive anomaly in SSP5-8.5 compared to SSP1-2.6. The color bars in (c) and (d) apply to both subfigures