Past climate variability and extreme events cannot be understood from any single source of information. Climate-model simulations provide physically consistent representations of past climate states and the processes governing variability and extremes, but they cannot reproduce the unique sequence of events that actually occurred. Geological and biological proxies, historical documents, archaeological records and other event-based archives instead preserve evidence of the realised past, but with heterogeneous temporal resolution, spatial coverage, sensitivity, preservation and uncertainty. This session invites contributions that combine, compare or jointly interpret multiple sources of information to investigate past climate change, variability and extremes.

We particularly welcome studies integrating climate-model simulations with geological, palaeoenvironmental, documentary, historical or archaeological evidence; approaches addressing rare or high-impact events such as storms, floods, droughts and heat extremes; statistical or probabilistic methods designed to account explicitly for dating uncertainty, proxy sensitivity, observational thresholds, preservation biases and model uncertainty.

Of particular interest are approaches that move beyond simple model–data comparison towards a process-based understanding of what different sources can reveal about past climate states and events. Contributions exploring data assimilation, probabilistic inference, multi-proxy synthesis, event attribution, high-resolution palaeoclimate modelling, machine learning, or methods for identifying convergence and disagreement among heterogeneous evidence are encouraged.

The session aims to foster dialogue across traditionally separate communities and to explore how complementary information from models and archives can be combined to produce more physically grounded and uncertainty-aware reconstructions of past climate variability and extremes.