Considerable Uncertainty of Simulated Arctic Temperature Change in the Mid-Holocene Due To Initial Ocean Perturbation

Arctic temperature is one of the most uncertain aspects of mid-Holocene (MH) climate change modeling, usually attributed to the different responses of different models to external forcing. However, in this study, we find that significant discrepancies (i.e., the noise is close to the signal in term of climate change) in the MH Arctic temperature changes can occur within the same model and for identical external forcing due to initial ocean condition perturbations. It is shown that initial ocean perturbations can affect the surface energy budget change through the uncertain cloud effect on shortwave radiation in boreal summer. The resulted uncertain change in summer surface heat flux alters the subsequent autumn and winter sea ice and contributes to significant differences in Arctic temperature via sea ice-albedo feedback. This study suggests that internal uncertainty of an individual model is a non-negligible source of overall uncertainty in simulating the MH Arctic temperature change. Proxies suggest that during the mid-Holocene (MH) (similar to 6,000 years ago), the climate was warmer than in the preindustrial period. However, climate models generally suggested a colder MH climate. This discrepancy between data and model results could be attributed to either the biased interpretation of proxy records or the models' unrealistic response to external forcings. From the modeling perspective, the Arctic temperature change is one of the most uncertain aspects of MH simulation, which can hinder our understanding of MH climate change and the comparison with proxy data. Generally, the uncertainty in modeling Arctic temperatures is attributed to variations in the response to external forcings among different models. However, we have uncovered that this uncertainty also exists within a single model. The responses of Arctic temperatures differ significantly in identical experiments when different initial ocean states are used. Specifically, the influence of clouds on the climate is highly sensitive to the initial ocean conditions, leading to various effects on solar radiation, which in turn affects changes in Arctic sea ice to varying degrees, ultimately resulting in distinct Arctic temperature changes during the MH. This study underscores the importance of addressing internal uncertainties within individual climate models when simulating changes in Arctic temperatures. The Arctic temperature changes in the mid-Holocene (MH) exhibit considerable discrepancy among the experiments with initial ocean perturbation The summer cloud effect on solar insolation plays a crucial role in modifying the Arctic sea ice change and hence the temperature response The internal uncertainty within a single model is non-negligible in simulating the MH Arctic temperature change