MODEL DESCRIPTION, PARAMETRIC CONFIGURATIONS & ANALYSIS DATA: Linking Atmosphere-Only and Coupled Climate Simulations Through Perturbed Parameter Ensembles

Wave45 Parametric Configuration – IPSL-CM6.5.2 PPE Ensemble Publication associated with this dataset is currently under review in JAMES Linking Atmosphere-Only and Coupled Climate Simulations Through Perturbed Parameter Ensembles B. S. Ferster¹˒²†, J. Mignot², G. Gastineau², F. Hourdin³, M. Coulon–Decorzens³, J. Deshayes², G. Gachon², J. Puyssegur² ¹ CNRM, Université de Toulouse, Météo France, CNRS, Toulouse, France² LOCEAN-IPSL (Sorbonne Université, CNRS, IRD, MNHN), Paris, France³ LMD-IPSL, Sorbonne University, CNRS, Paris, France Dataset description This dataset contains the parameter configurations and selected analysis-ready output associated with the Wave45 perturbed parameter ensemble (PPE) conducted with the IPSL coupled climate model IPSL-CM6.5.2. The model is derived from the CMIP6 IPSL-CM6A-LR configuration (Boucher et al., 2020), with updated ocean and sea-ice components and minor bug corrections. The ensemble explores structural and parametric uncertainty in coupled climate simulations using a constrained subset of physically relevant parameters obtained through iterative history matching. Model configuration The coupled model consists of the following components: Atmosphere: LMDZ6 (Hourdin et al., 2020) Land surface: ORCHIDEE (Krinner et al., 2005) Ocean: NEMO v4.2 (Madec et al., 2022) Sea ice: SI3 (Vancoppenolle et al., 2023) The atmospheric grid has a horizontal resolution of 144 × 143 points (approximately 2.5° × 1.25°) with 79 vertical levels extending to approximately 80 km. The ocean uses the ORCA1 tripolar grid (approximately 1° nominal resolution) with 75 vertical levels. The land model shares the atmospheric grid and includes 11 hydrological soil layers extending to 2 m, with a deeper thermal diffusion extension to 90 m. All simulations are forced with fixed CMIP6 year-2000 greenhouse gas and aerosol concentrations. Experimental design The Wave45 ensemble forms part of a perturbed parameter ensemble framework constructed using iterative history matching and refocusing, implemented with HighTune Explorer and building on the methodology described by Hourdin et al. (2023). The full parameter exploration uses 12 free parameters spanning atmospheric, land, ocean, and sea-ice processes (P1–P12). These parameters were selected based on their expected influence on climate sensitivity, radiative balance, and atmosphere–ocean coupling. Wave45 selection Wave45 refers to the 45th and final wave of the history-matching procedure. The final ensemble consists of 120 parameter vectors randomly sampled from the final Not Ruled Out Yet (NROY) parameter space after the iterative history-matching procedure. The 120 parameter vectors define the final constrained PPE and are provided in Wave_45_Params.nc. The parameter configurations use the same 12-dimensional parameter definition for the atmosphere-only and coupled model configurations. The complete parameter/configuration files required to define the PPEs are provided in the companion Zenodo dataset cited below. Reproducibility To reproduce the model simulations, users should: Obtain the IPSL-CM6.5.2 model code and associated LMDZ6–ORCHIDEE–NEMO–SI3 coupled system. Apply the 12-dimensional parameter vectors provided in the archived parameter/configuration files. Run the corresponding atmosphere-only and coupled simulations using the model configurations described above. The history-matching procedure and parameter selection follow the iterative methodology described in Hourdin et al. (2023). Additional information on the history-matching workflow and the final NROY parameter space is provided in the analysis data archived here. Data access The complete model output from the history-matching procedure and the final PPE is retained on IPSL computing infrastructure because its total size exceeds the storage limit of this repository. The present archive therefore provides a selected, analysis-ready subset of model output together with the derived data and analysis software required to reproduce the principal analyses and quantitative results presented in the manuscript. Analysis software and supporting data To facilitate reproducibility of the analyses presented in the manuscript, this dataset is complemented by analysis notebooks, processed model output, observational data, and derived products used to produce the principal results and figures. HighTune version The history-matching procedure was performed using HighTune. The exact version used was revision 397 of the HighTune trunk: svn checkout --username=htune https://svn.lmd.jussieu.fr/HighTune/trunk -r 397 IPSL0 Checked out revision 397. The archived files associated with the history-matching procedure include the observational targets and remaining parameter-space information obtained throughout waves 1–45. Analysis notebooks The following Jupyter notebooks are provided: Figure_1.ipynb: analysis and plotting associated with Figure 1. Compute_EOF_Example.ipynb: example calculation of the EOF decomposition and principal components, including the calculation of the SST-removed fields. Figure_2.ipynb: analysis and plotting associated with Figure 2. NROY_Waves_Plot.ipynb: visualization of the evolution of the remaining NROY parameter space over history-matching waves 1–45. Metrics_Plot.ipynb: visualization of the climate metrics used in the history-matching analysis. The notebooks are provided together with the processed input data required for the corresponding calculations. History-matching data The following files document the parameter constraints and history-matching procedure: Observed_Metrics_REF_1D_3D.csv: observational/reference targets used for the 1D and 3D atmosphere-only history-matching stages. Observed_3D_Metrics_HighTune.nc: observational/reference metrics used by the 3D history-matching procedure. Wave_45_Params.nc: parameter values of the final 120-member parameter ensemble. LMDZ_Remaining_space_waves_1_45.nc: remaining/NROY parameter-space information from waves 1–45 for the atmosphere-only history-matching procedure. IPSL_CM6_Remaining_space_waves_45.nc: remaining/NROY parameter-space information for the final coupled configuration. metrics/: processed climate metrics used during the history-matching procedure. These files document the progression of the constrained parameter space and the final parameter ensemble used to construct the atmosphere-only and coupled PPEs. EOF and SST-removal analysis The EOF analysis and removal of the component linearly associated with ensemble variation in mean SST are supported by the following processed data: IPSL-CM6-W45-Corrected-temp_ANM_Spatial_*.nc LMDZ-W45-temp_ANM_Spatial_*.nc IPSL-CM6-W45-Corrected_temp_weighted_sst_ANM_Spatial_*.nc IPSL-CM6-W45-Corrected-temp_weighted_sst_regression.nc The files containing EOF, PC, Cov, Eval, and Var correspond to the EOF decomposition and associated diagnostics. The SST-related files contain the regression and residual fields used to separate the component of ensemble variability linearly associated with the 60°S–60°N mean SST. Compute_EOF_Example.ipynb provides an example of how these quantities are calculated from the archived model data. Atmospheric, radiation, ocean, and sea-ice diagnostics The archive also contains analysis-ready subsets of model output used to investigate the atmospheric, radiative, ocean, and sea-ice responses of the PPEs. These include: IPSL-CM6-W45-temp-zonalmean-ANM.nc LMDZ-W45-temp-zonalmean-ANM.nc IPSL-CM6-W45-tos-globalmean-ANM.nc IPSL-CM6-W45-TOA-metrics-ANM.nc LMDZ-W45-TOA-metrics-ANM.nc IPSL-CM6-W45-cloudiness-globalmean-ANM.nc LMDZ-W45-cloudiness-globalmean-ANM.nc IPSL-CM6-W45-rneb_geoph_temp_vitu-zonalmean-ANM.nc LMDZ-W45-rneb_geoph_temp_vitu-zonalmean-ANM.nc IPSL-CM6-W45-siconc_JFM_Spatial.nc IPSL-CM6-W45-siconc_JAS_Spatial.nc LMDZ-W45-ATM_variables-Spatial-ANM.nc IPSL-CM6-W45-ATM_variables-Spatial-ANM.nc LMDZ_Area_Mask.nc These files provide the processed atmospheric temperature, TOA radiation budget, cloudiness, SST, and seasonal sea-ice diagnostics required for the manuscript analyses and figures. Observational data ERSSTv5_globalmean_sst.nc contains the ERSSTv5 global-mean SST data used for comparison with the model results. The archived observational data cover 1980–2014. Huang, B., Thorne, P. W., Banzon, V. F., Boyer, T., Chepurin, G., Lawrimore, J. H., Menne, M. J., Smith, T. M., Vose, R., & Zhang, H.-M. (2017). Extended Reconstructed Sea Surface Temperature, Version 5 (ERSSTv5): Upgrades, Validations, and Intercomparisons. Journal of Climate, 30, 8179–8205. https://doi.org/10.1175/JCLI-D-16-0836.1 Scope of the archived model output The archived model output is a selected, analysis-ready subset of the complete simulations. It has been designed to provide the data required to reproduce the principal analyses, quantitative results, and figures presented in the manuscript without requiring access to the complete model output. The complete model output from the history-matching procedure, including approximately 600 atmosphere-only simulations and the full coupled PPE output, is not included because of its size. The complete simulations are retained on IPSL computing infrastructure. The archive therefore distinguishes between: the parameter and configuration files required to define the PPEs; the selected model output required for the manuscript analyses; and the derived data products and analysis notebooks used to obtain the published results. The remaining data required to reproduce the figures and analyses presented in the corresponding Ferster et al. manuscript are provided in this archive. These include atmospheric temperature diagnostics, top-of-atmosphere (TOA) radiation budget diagnostics, and seasonal sea-ice concentration data. The procedure used to obtain the SST-removed fields is documented in Compute_EOF_Example.ipynb, which provides an example of the EOF decomposition and the ensemble-based removal of the component linearly associated with mean SST. The accompanying analysis-ready NetCDF files contain the derived EOFs, principal components, covariance fields, and related diagnostics used in the manuscript. Contact For questions or requests: [e-mail]