Interannual to Multidecadal Climatic Fluctuations Due to Volcanic Eruptions

Low-frequency (decadal to centennial) modes of ocean variability are important components of climate variability. These modes are often inferred from long-term climate simulations after being preprocessed by low-pass filtering. Notably, the few modes that are consistently found in many climate models differ significantly, even in frequency, as every model has biases and model errors. At the same time, validation of the extracted signals against observations is limited by the time span of the observational record (e.g., sea surface temperature and sea ice extent observed during the satellite era), which is often shorter than the timescales of interest and also significantly altered by anthropogenic factors. More importantly, due to preprocessing as well as the subsequent data analysis techniques, such as empirical orthogonal functions (EOFs), the results have frequently ambiguous physical interpretation.

Here, we investigate interannual to multidecadal response of oceanic circulation to volcanic eruptions and their possible impact on climate (e.g., droughts, glaciations). We also study other modes of coupled atmosphere-ocean climate variability (e.g., North Atlantic Oscillation) and establish linkages between these patterns and low-frequency perturbations of climate triggered by volcanic eruptions. For that, we study the Last Millennium Ensemble (https://www2.cesm.ucar.edu/models/experiments/LME). We analyze individually over 70 eruptions of both tropical and extratropical origin, with a particular focus on their regional impact (e.g., Baffin Island, Sahel, Australia) on long-term perturbation of precipitation or ice cover. We capture patterns of coupled atmosphere-ocean response to volcanic eruptions through multi-component composites, providing physical mechanisms behind regional climatic response to volcanic eruptions. Moreover, we apply a recently introduced technique called Nonlinear Laplacian Spectral Analysis. Through this technique, drawbacks associated with ad-hoc filtering are avoided as the extracted signals span many temporal scales without preprocessing the input data, enabling detection of low-frequency, low-amplitude and intermittent modes otherwise not accessible with classical approaches.