Predictability and Future Projections of Weather Conditions Relevant to COVID-19 Mortality

During the early stages of the COVID-19 pandemic, we proposed a relationship between COVID-19 mortality and specific temperature and humidity ranges (Sajadi et al., 2020; http://dx.doi.org/10.2139/ssrn.3550308). Our analyses had predictive power and foresaw the second wave of mortality that swept the CONUS in the autumn of 2020.

In this paper, we first re-examine our hypothesis using the most up-to-date data of (1) weekly mortality at the state level and the quantiles of SARS-CoV-2 density in wastewater provided by the CDC, and (2) meteorological data from the ECMWF ERA5 reanalysis. We examine differences in mortality between northern and southern states and suggest a new hypothesis explaining their distinct periodicities. We then explore the predictability of the proposed criterion using reforecasts conducted with NOAA’s GEFSv12. We show that the conditions beneficial to SARS-CoV-2 transmission are predictable at least two weeks in advance. Finally, we investigate future changes in the statistics of atmospheric conditions that are beneficial to the transmission of SARS-CoV-2 using data from the CMIP6 experiment. We conclude by discussing the importance of including information on atmospheric conditions in predictive epidemiological models.