Probabilistic Storylines: Conditional Likelihoods of the 1997 California New Year’s Flood Under Global Warming Levels

The 1997 New Year's Flood was the most costly flood in California history, a compound extreme event driven by a category 5 atmospheric river carrying extreme precipitation, and amplified by snowmelt and elevated antecedent soil moisture. Previous work has successfully recreated the event using regionally-refined model meshes, identifying the major drivers of the flood, demonstrating the importance of model horizontal resolution in representing runoff totals, and demonstrating the warming sensitivity of these flood drivers. However, analyses have stopped short of estimating the likelihood of a comparable flood occurring under future climate conditions. Estimating the likelihood of single variable, much less compound extremes, under future climate is challenging, due to the multiplicity of uncertainty sources, the challenges in navigating deep uncertainties, and the limitations of comparable large ensembles and simulation capabilities between climate models. Building on recent work proposing a new conceptual methodology for "probabilistic storylines", the work presented here addresses this gap by estimating conditional probabilities for the 1997 flood storyline using the Energy Exascale Earth System Model (E3SM). Univariate and multivariate thresholds are calculated using the E3SM reanalysis, and return likelihoods and risk ratios under 1.5°C, 2°C, and 3°C global warming levels are estimated using the E3SM historical and SSP370 large ensemble. Multivariate joint probabilities of compound flood drivers are evaluated using copulas. Results quantify how the likelihood of a 1997 flood-like compound extreme evolves under different warming levels, conditional on the global climate sensitivity and regional structural uncertainty of E3SM.