Abstract Ambient temperature is an established health determinant, and having its data representation at high spatial resolution is critical for understanding the climate‐health nexus. Here, we compare temperature data sets for modeling the association between ambient temperature and emergency department (ED) visit related to diarrhea, a disease with well‐documented temperature sensitivity. We evaluate six temperature data sets: one from in situ weather stations, three reanalysis products (ERA5 surface and air temperatures, GridMET), and two temperature data sets derived from satellite observations (GOES‐16, Landsat). The study was conducted over New York City where daily ED data at the ZIP code level was available. Using the distributed lag nonlinear model, our analysis shows that at the aggregated city level, these data sets produced similar results and were consistent with the one modeled through weather station data that is the current golden standard. Air temperature and surface temperature were consistent with each other when processed as percentiles. Increasing risk was associated with increasing temperature, with mean relative risk of 1.21–1.26 (95% empirical CI 1.14–1.34) at the 90th percentile depending on data compared with at the 50th percentile. Older adults aged 65+ were more vulnerable (1.49–1.59). High‐resolution data was better in capturing temperature variability between ZIP codes and allowed for estimating relative risk individually. The findings from this study support using high‐resolution data for more accurate temperature assessment in modeling the climate‐health nexus.
Liu et al. (Mon,) studied this question.