Relating synoptic air masses to human mortality

Prior research has studied the utility of various metrics of thermal comfort in an effort to elucidate the nuanced relationship between temperature and human mortality. Synoptic-scale air masses (AMs) – sometimes referred to as surface weather types – are one of these metrics. Within the last 5 years, a new global-scale AM classification dataset was published – the GWTC2 – which has not yet been evaluated in terms of its relationship to human mortality. The GWTC2 is a system of multivariate surface AM classification, identifying spatiotemporally-relative AMs, and frontal passages, for every day (from 1979-present) at every location (at 0.5° x 0.5° spatial resolution) across the globe. Thus, this research leverages the GWTC2 to examine the lagged relationship between AMs and anomalous mortality for 61 cities across various climates zones of the continental United States. Results show that AMs are significantly related to human mortality, at most locations and in all seasons of the year. The Humid-Warm (HW) and Warm (W) AMs yielded significant and immediate (at 0-1 day lag times) increases in mortality in summer, though the immediate increases after HW occurred were not followed by decreases in mortality – indicating a lack of the “mortality displacement” effect highlighted in some prior research. Meanwhile, the Dry-Cool (DC) and Cool (C) AMs both resulted in a delayed, but extended period (from 1 to 19 days of lag) of excess mortality in all seasons outside of summer, generally peaking 2 to 4 days after the occurrence of the DC or C air mass. Of note, warm front passages (increased mortality) and cold front passages (decreased mortality) yielded significant anomalous mortality responses one day after their occurrence in every season of the year – the most seasonally-consistent AM-mortality relationship discovered. Furthermore, many of these results were consistent across the majority of the 61 cities examined. Finally, artificial neural network (ANN)-based modeling revealed that a meta-model (i.e., a model developed on the combined data of all 61 locations) was more skillful in predicting AM-mortality relationships for many cities than developing 61 individual ANN models for each city. Future research should explore whether the consistency of these results holds up when exploring the impact of GWTC2 air masses in other parts of the globe.