We address this question by first identifying major heat events, then using machine learning to classify their associated flow patterns, and finally performing back trajectories to determine the sources of air and thus the physical mechanisms responsible for the extreme heat.
First we identify heat waves in the ERA5 reanalysis using a set of criteria including a combination of relative and absolute thresholds, heat index and temperature fields, minimum spatial extent, and minimum duration. The goal of this identification algorithm is to objectively select only strong, large-scale, and long-lived heat waves rather than transient heat events which may have minimal impacts. We then use self-organizing maps to objectively classify flow patterns associated with heat waves by training on middle-tropospheric heights and wind velocities. The LAGRANTO software (Sprenger and Wernli 2015) is used to perform back trajectories for the cases in each cluster and quantify the contribution of different heating mechanisms– advection, subsidence, condensational heating–as per Röthlisberger and Papritz (2023). We also investigate regional differences in heat wave characteristics such as size, duration, and magnitude, as well as the processes that drive them.
References:
Röthlisberger, M., Papritz, L. (2023). Quantifying the physical processes leading to atmospheric hot extremes at a global scale. Nat. Geosci. 16, 210–216, https://doi.org/10.1038/s41561-023-01126-1
Sprenger, M. and Wernli, H. (2015). The LAGRANTO Lagrangian analysis tool – version 2.0, Geosci. Model Dev., 8, 2569–2586, https://doi.org/10.5194/gmd-8-2569-2015

