Feasibility of climate reanalysis data as a proxy for onsite weather measurements in outdoor thermal comfort surveys.

Outdoor thermal comfort (OTC) surveys require synchronous monitoring of meteorological variables for direct comparisons against subjective thermal perception. The Universal Thermal Climate Index (UTCI) is a feasible index as it integrates meteorological conditions as a single value irrespective of urban morphological attributes or biological sex, age and body mass. ERA5-HEAT (Human thErmAl comforT) is a downloadable reanalysis dataset providing hourly grids of UTCI climate records at 0.25° × 0.25° spatial resolution from 1979 to present. We here evaluate for the first time whether it is possible to use ERA5-HEAT data as a proxy for the UTCI measured onsite during OTC surveys. A dataset comprising 1640 survey responses gathered over 14 OTC campaigns in Curitiba, Brazil (25°26'S, 49°16'W) was analysed. We assessed the bias obtained between the Dynamic Thermal Sensation, an index derived from the UTCI, and the thermal sensation reported by survey participants by considering locally measured meteorological variables and ERA5-HEAT reanalysis data. As ERA5-HEAT data are given on an hourly basis, prediction bias can be greatly reduced when accounting for survey responses close to the hour. In terms of seasons, the fall and winter seasons have diminished mean bias, though with larger spread than in summer. In terms of UTCI stress categories, prediction bias is lower for the thermal comfort range. When comparing reanalysis data against WMO station data as proxy candidates for survey field data, the former presented lower bias, less spread in terms of standard deviation and higher correlation to