Resumen
An increasing number of people living in urban environments and the expected increase in long lasting heat waves makes the study of temperature distribution one of the major tasks in urban climatology, especially considering human health and heat stress. This excess heat is often underestimated because stations from national meteorological services are limited in numbers and are not representing the entire urban area with typically higher nocturnal temperatures, especially in densely built-up environments. For a majority of the population, heat stress is consequently monitored insufficiently. In this study, the factors influencing the nocturnal urban heat island have been evaluated in detail and have been tested using different spatial resolutions. A multiple linear regression model has been developed with predictors resulting from different data sources to model the urban air temperature distribution continuously. Results show that various datasets can be used for the prediction of the heat island distribution with comparable results, ideally run on a 200 m grid. Validation using random sampling indicated a RMSE clearly below the standard deviation of the measurements with an average around ~0.15 °C. The regression coefficients are varying within the nocturnal runs with best results around 22:00 CET (R2 > 0.9).