Crowdsourcing air temperature from citizen weather stations for urban climate research

•Novel data set for monitoring and analyses of urban air temperatures•One-year data set from up to 1500 citizen weather stations in Berlin, Germany•Comprehensive quality assessment of crowdsourced air temperature (Tcrowd)•Quality-checked Tcrowd show spatio-temporal patterns of the urban heat island•High spatial density of Tcrowd in Berlin and other cities

Provision of accurate air temperature data in urban environments with high spatial and temporal resolution over long time periods remains a challenge in atmospheric research. Crowdsourcing, i.e., collection of atmospheric data from non-traditional sources like citizen weather stations (CWS), is an alternative and cost-efficient method for exploration and monitoring of urban climates. This study examines the suitability of crowdsourced air temperature (Tcrowd) measurements from CWS by comparing Tcrowd from up to 1500 stations with reference air temperature (Tref) in Berlin and surroundings for a period of twelve months (Jan-Dec 2015).Comprehensive quality assessment of Tcrowd reveals that erroneous metadata, failure of data collection, and unsuitable exposure of sensors lead to a reduction of data availability by 53%. Spatially aggregated raw data of Tcrowd already provide a robust estimate of hourly and daily urban air temperature in the study area. Quality-checked Tcrowd observations show spatio-temporal characteristics of the urban heat island in Berlin with higher spatial variability than Tref in built-up areas. Spatial density of Tcrowd in Berlin exceeds that of the reference monitoring network by far. However, rigorous data quality assessment is the key challenge in order to fully benefit from this novel data set for urban climate research.