Comparison of temperature-mortality associations using observed weather station and reanalysis data in 52 Spanish cities

Abstract

Background: Most studies use temperature observation data from weather stations near the analyzed region or city as the reference point for the exposure-response association. Climatic reanalysis data sets have already been used for climate studies, but are not yet used routinely in environmental epidemiology.

Methods: We compared the mortality-temperature association using weather station temperature and ERA-5 reanalysis data for the 52 provincial capital cities in Spain, using time-series regression with distributed lag non-linear models.

Results: The shape of temperature distribution is very close between the weather station and ERA-5 reanalysis data (correlation from 0.90 to 0.99). The overall cumulative exposure-response curves are very similar in their shape and risks estimates for cold and heat effects, although risk estimates for ERA-5 were slightly lower than for weather station temperature.

Conclusions: Reanalysis data allow the estimation of the health effects of temperature, even in areas located far from weather stations or without any available.

Keywords: Distributed lag non-linear models; Mortality; Reanalysis; Spain; Temperature; Weather station.