Quantifying the tuning uncertainty on the climate sensitivity of the EC-Earth climate model

Despite decades of developments, the inter-model spread in climate sensitivity in the latest CMIP ensemble remains substantial, with relevant implications for mid- to long-term climate projections. The inter-model differences are driven by model biases and structural deficiencies, mostly linked to cloud feedbacks, but the specific processes that dominate this issue remain unclear. Physical parametrizations are of primary importance for the performance of climate models, in particular those regarding microphysics and convection - especially at current model resolutions. Indeed, a fundamental yet often overlooked aspect of coupled model development is the tuning of parameters involved in these parametrizations to align with some specific constraints (e.g. radiative balance and global mean temperature in the pre-industrial state).
Here, we propose a methodology to evaluate the uncertainty in equilibrium climate sensitivity (ECS) arising from parameter tuning and apply it to the EC-Earth3 climate model. Our approach consists in systematically perturbing a set of tuning parameters - primarily those affecting tropical convection and precipitation - aiming to maximize their impact on climate sensitivity while ensuring the parameters remain within a plausible range. We obtain a low and a high sensitivity configuration of the model, resulting in a moderate change in climate sensitivity of approximately ±0.3 K. Finally, the results are discussed in the context of the CMIP6 ensemble, suggesting that the inter-model spread is likely driven by deeper structural differences within the models rather than uncertainties arising from the tuning process.