Abstract:
Human activities are increasingly leading to the emission of greenhouse gases, altering the Earth’s climate into an unprecedentedly warmer state, thus compromising our ability to devise effective adaptation strategies to climate change impacts. Although paleoclimates are not perfect analogues for these warming trends, understanding past climate dynamics provides valuable insights into future climate change. These past climates span a tremendous range of hydroclimates, landscapes, and biodiversity distributions that can contribute to our understanding of the key elements of the climate system and also serve as out-of-sample validations for the strength and stability of climate sensitivity and feedbacks in climate models to ensure accurate future projections. However, understanding the past is contingent upon the availability and accurate interpretation of climate signals from paleoclimate records. Stable isotope ratios of oxygen (18O/16O; δ18O) and hydrogen (2H/1H; δ2H) in water imprints in the hydrological cycle reflect many integrated processes of the Earth’s system and form the basis of paleoclimate reconstruction. The interpretation of the isotopic composition of precipitation (δ18Op) signals from paleoclimate records faces significant challenges and uncertainties due to the wide range of large-scale and local climatic and environmental conditions that control its spatio-temporal variability. This implies that the δ18Op signal requires the disentangling of climate signals from non-climate signals and needs paleoclimate-constrained transfer functions to ensure accurate interpretations. This thesis demonstrates how isotope-enabled General Circulation Models (iGCMs) can be combined with paleoclimate records to enhance the interpretability of paleoclimate dynamics. Part 1 integrates iGCMs with stable isotope paleoaltimetry to reconstruct the Miocene Central Alps paleoelevation. Through topographic sensitivity and Middle Miocene climate experiments, the results show that using contemporary isotopic lapse rates overestimates the paleoelevation by ~1.5 km, suggesting the need for refining the previous estimates with iGCM-simulated paleoclimate-constrained isotopic lapse rates. Part 2 presents an extensive suite of (paleo)climate experiments with iGCMs from present-day to Mid-Pliocene conditions to understand how large-scale atmospheric modes of variability (i.e., North Atlantic Oscillation and East Atlantic Oscillation patterns) and West African monsoon dynamics influence the regional hydroclimate and δ18Op patterns across Europe and West Africa. Through statistical analysis (e.g., correlations and causality testing), the results indicate that the causal links between the local isotopic proxy and large-scale patterns and regional hydroclimate variables are significantly different under the varied past climates. This proposes the need to understand the time and space-dependent relations between proxy systems and regional paleoclimate dynamics to refine their transfer functions. Due to the computational cost of the proposed paleoclimate reconstruction framework, Part 3 further explores the potential of using machine learning to emulate the spatio-temporal variability of δ18Op values. The results indicated overall good performance that was at least better than iGCM. Altogether, the findings indicate the importance of combining water isotopologue information from observations, iGCMs, and isotopic paleoclimate records to provide robust statistical and dynamical constraints on paleoclimate reconstructions, which has huge implications for reducing the uncertainties of climate models and thus improving future climate projections.
Paleoaltimetry