A High-resolution Global-scale Model for COVID-19 Infection Rate

Abstract

Several models have correlated COVID-19 spread with specific climatic, geophysical, and air pollution conditions, and early models had predicted the lowering of infection cases in Summer 2020. These approaches have been criticized for their coarse assumptions and because they could produce biases if used without considering dynamic factors such as human mobility and interaction. However, human mobility and interaction models alone have not been able to suggest more innovative recommendations than simple social distancing and lockdown, and would definitely need to include information about the base environmental suitability of a World area to COVID-19 spread. This scenario would benefit from a global-scale high-resolution environmental model that could be coupled with dynamic models for large-scale and regional analyses. This article presents a 0.1 high-resolution global-scale probability map of low and high-infection-rates of COVID-19 that uses annual-average surface air temperature, precipitation, and CO2 as environmental parameters, and Italian provinces as training locations. A risk index calculated on this map correctly identifies 87% of the World countries that reported high infection rates in 2020 and 80% of the low and high infection-rate countries overall. Our model is meant to be used as an additional factor in other models for monthly weather and human mobility. It estimates the base environmental inertia that a geographical place opposes to COVID-19 when mobility restrictions are not in place and can support how much the monthly weather favors or penalizes infection increase. Its high resolution and extent make it consistently usable in global and regional-scale analyses, also thanks to the availability of our results as FAIR data and software as an Open Science-oriented Web service.