State-dependent effects of natural forcing on global and local climate variability

Climate variability is the primary influence on climate extremes and affected by natural forcing from solar irradiance and volcanic eruptions. Global warming impacts climate variability, but there is contradictory and incomplete evidence on the spatio-temporal patterns. Strong volcanic eruptions have been suggested to reduce temperatures less in warmer climate states. However, the underlying question of state-dependent effect of natural forcing on local and global variability remains open. Moreover, there are uncertainties about the role of natural forcing in the mismatch between simulated and reconstructed local, long-term variability.  

Using a 12-member GCM ensemble with targeted boundary conditions, we present naturally-forced and equilibrium, millennium-length simulations for the Last Glacial Maximum (LGM) and the Pre-Industrial (PI). We quantify the local and global climate response to solar and volcanic forcing in the LGM and PI, and contrast variability from forced and control simulations on annual-to-multicentennial scales. We differentiate various contributions from the atmosphere, oceans, and particularly that of sea ice using a 2D energy balance model (EBM). Spectral analysis of simulated temperatures shows that global variability is predominately determined by natural forcing. Local mean spectra are more characteristic for the mean climate state and reveal a decrease in local variability with warming. The global and local response to natural forcing is robust against changes in the mean climate. Particularly, the spatial patterns of the surface climate's response to volcanic eruptions widely agree across states. Weak local differences resulted primarily from sea ice dynamics. The sea ice contribution is the strongest on interannual scales. It remains significant on decadal scales and longer, providing a key mechanism of long-term variability. We validate the simulated variability against observational and paleoclimate data. The variance obtained from proxies is increasingly larger on longer timescales compared to that from simulated time series. The inclusion of natural forcing reduces the model-data mismatch on decadal-to-multicentennial scales and, thus, provides a more accurate representation of climate variability. 

Consideration of natural forcing is therefore paramount for model-data comparison and future projections. The robust temperature response suggests that findings on the ability of models to simulate past variability should translate to future climates, and can thus help constrain variability.