The results of atmospheric nudging experiments over the Labrador Sea in the CAM6-POP2 model using MERRA-2, and source code related to the main Figures

Simulating the AMOC response to surface heat flux forcing Statistical analyses of CMIP6 multi-model simulations suggest the role of anomalous ocean heat uptake as an initiating factor in future AMOC weakening, followed by reduced salt transport that can further weaken the AMOC. The analysis indicates that the extent of anomalous heat uptake in the Labrador Sea is likely one of the major contributors to inter-model differences in future AMOC strength. To verify these processes, we conducted idealized sensitivity experiments using CESM2 atmosphere-ocean coupled configurations. Unlike the CMIP6 correlation analysis, the perturbation experiments offer direct mechanistic evidence for causality, because the simulated AMOC weakening arises exclusively from the imposed Labrador Sea heat flux forcing. This controlled framework enables us to isolate and quantify the causal influence of surface heat flux anomalies on AMOC variability, rather than reproducing the CMIP6 multi-model mean response or its inter-model spread. a. Idealized model experimentsWe conducted experiments including control simulations and forced experiments, targeting the Labrador Sea region. Each experiment consists of three ensemble members with different initial conditions. The aim was to isolate the effects of modified atmospheric forcing on the AMOC in this region.1) Control (Labrador Sea): Over the Labrador Sea region (50°N–75°N, 45°W–70°W), daily climatologies (1980–2014) of U, V, Ta, and qa from MERRA-2 were restored from 1980 to 2030.2) Heat Flux Forcing (Labrador Sea): To examine the AMOC’s sensitivity to the increased NSHF, we conducted a nudging experiment using the same setup as the Control (Labrador Sea) experiment. The same restoration procedure was applied, but Ta and qa were increased while U and V were reduced over the Labrador Sea region to generate enhanced downward NSHF anomalies. Specifically, U and V were decreased by 30% to weaken surface wind stress, while Ta and qa were increased by 20%. In the control mean state, this 20% scaling corresponds to perturbation amplitudes of ΔT = 4.4°C and Δq = 0.87 g kg-1, respectively. File name structure --source code for main Figures and colormap Draw_Fig{num}.ncl (source code for main figures) For_Figure{num}.nc (data for source code) {num}: number of main Figures bgwyr.rgb: ncl color map --model results {Forcing_location}_{EXP}_{ENSNUMB}_{VAR}_1980_2030.nc Forcing_location: Labrador Sea region (Labrador) EXP: Control and nudging experiment (Control, Forcing) ENSNUMB: Ensemble member indices (ens0, ens1, ens2) VAR: variables (AMOC26N, HT50N, ST50N, SST, SSS, SHF, TEMP0_1000m, SALT0_1000m) ========================================================================== Variable list of model results AMOC26N: AMOC is calcuated using meridional ocean velocity at 26°NHT50N: Heat transport is calculated using meridional ocean velocity and ocean temperature at 50°NST50N: Salt transport is calculated using meridional ocean velocity and ocean salinity at 50°NSST: Sea surface tempeartureSSS: Sea surface salinitySHF: Net surfcae heat flux (positive downward)TEMP0_1000m: Temperature (0-1000m depth weighted averaged)SALT0_1000m: Salinity (0-1000m depth weighted averaged)========================================================================== ========================================================================== note CAM6 nudging experiments were conducted by forcing the CAM6 model with MERRA-2 atmospheric data (bottom temperature, humidity, and wind) over the Labrador and Irminger Seas. CTRL (Control simulation): Nudging with climatological values. Forcing (Atmospheric nudging simulation): Restores atmospheric conditions to warmer temperatures, higher humidity, and slower winds over the Labrador Sea region. Access informationNudging data (MERRA-2) for model simulation is obtained from the following sources: https://rda.ucar.edu/datasets/d313003/.