Warming climates across timescales: insights from the climate model AWI-CM3

The Earth’s recent, rapid shift from a naturally forced icehouse climate to an anthropogenically driven hothouse climate has sparked the urgency to understand past warm climates to constrain future climate model simulations. In this sense, three past warm climates whose forcings are relatively well constrained with respect to Pre-Industrial (PI; 1850 Common Era (CE)) conditions, provide a good testbed for climate models. These include the mid-Holocene (MH; 6 kiloyears (ka) before present (BP)), the Last Interglacial (LIG; 127 ka BP) and the Late Pliocene (LP; 3.205 million years (Ma) BP). In this study, the large-scale climate features of these three periods are quantified through a series of core and sensitivity experiments that integrate the protocols of the Coupled Climate Model Intercomparison Project (MIP), phase 6 (CMIP6), the Paleoclimate MIP, phase 4 (PMIP4), and the Pliocene MIP, phase 3 (PlioMIP3), employing the Alfred Wegener Institute Climate Model version 3 (AWI-CM3). The focus is on Earth’s climate sensitivity to variations in Earth’s astronomical configuration, atmospheric greenhouse gas concentrations (GHGs), and paleogeography. To quantitatively and qualitatively describe the climate response to these forcings, eight time-slice quasi-equilibrated simulations were performed, four of which correspond to the core experiments covering the PI, LIG, MH, and LP periods. The remaining four represent targeted sensitivity experiments that address the PI climate with modified atmospheric CO2 concentrations (PI400 and 4xCO2), and the LP climate with modified orbital configuration (LP_highNH and LP_lowNH).
It is found that AWI-CM3 reproduces key features of the LP, LIG, and MH climates, particularly the spatial patterns of atmospheric and oceanic temperatures, precipitation, and ocean circulation, in line with proxy evidence and previous modelling efforts. Six regions are identified as hotspots of consistent climate response across the three warm periods, including the polar Northern Hemisphere, the Subpolar North Atlantic (SPNA) Gyre, the Mediterranean Sea, West Antarctica, the Bellingshausen-Amundsen sector of the Southern Ocean, the Indian Ocean, and the Amazon. These regions exhibit robust atmospheric and oceanic warming, as well as sea ice loss. The Amazon region exhibits drying in contrast to the wettening signal
of the other hotspots, while the SPNA gyre exhibits a saltier ocean in comparison to the remaining hotspots.
Although robust signs of the same response emerge at the near-surface across warm climates and timescales, the results presented in this dissertation reveal that ocean circulation is highly dependent on the model’s geography. This yields distinct circulation regimes across the ocean basins that are, to some degree, affected by orbital and CO2 forcing, but defined by the configuration of ocean gateways. In this sense, the LP-related experiments display a stronger Atlantic Meridional Overturning Circulation (AMOC), while increased CO2 concentration under modern geography promotes an AMOC weakening. No active Pacific MOC cell is found in any of the warm climate simulations, and all exhibit weakened Antarctic Bottom Water (AABW) export to the Atlantic while its formation in the Pacific is strengthened under modified orbital configurations.
This dissertation also addresses modelling challenges that have arisen during the design and execution of the climate simulations featured herein, as well as potential studies that can be derived directly from the available model output. Finally, the results presented in this dissertation demonstrate that the climate system exhibits consistent regional responses across different warm periods and forcing mechanisms, underscoring the value of past warm climates as natural laboratories for assessing Earth system sensitivity, improving model performance, and constraining projections of future anthropogenic warming.