Adapting weather files for urban microclimates using Personal Weather Station data

Abstract

Building simulation weather files are typically derived from long-term measurements at weather stations, which do not capture microclimatic variations at the neighbourhood scale where localized conditions may play an important role. The growth of Personal Weather Stations (PWS) has increased availability and coverage of weather data, providing opportunities to develop location or context-specific weather files. This paper describes a method for transforming standardized weather files using PWS data to capture climate variation due to tree canopy coverage in London, UK. PWS data is obtained from 490 Netatmo stations from 2017-2022, with quality control used to remove outliers. PWS are spatially joined to urban landcover data, including tree canopy coverage and built fraction, and temporally matched to weather data from Heathrow Airport, from which standard weather files are derived for London. A Generalised Additive Model is developed which predicts local air temperatures given Heathrow air temperature, landcover, month, and the time of day. The model is demonstrated by modifying Test Reference Year and Design Summer Year weather files for the 2020s and 2050s for different tree canopy coverages. The approach is reproducible and scalable, enabling customized weather files to be produced for locations where PWS and landcover data is available. The increasing density of PWS mean new opportunities for incorporating effects of urban climate into building simulation.