Investigating the impact of urban heat islands on long-term climatic observations in Sweden

Urban land cover (ULC) has been steadily expanding in Sweden over the last century. This expansion could potentially include areas in the vicinity of meteorological stations, and may, in turn, lead to increased urban heat island effects in the areas surrounding them. As observations form the basis of many climate studies, it is then important to investigate the potential influence of urban heat island effects on long-term trends in climatic observations. For the purpose of quantifying the change in ULC around meteorological stations, we developed a semi-supervised methodology that classifies the land cover based on single-band orthophotos, and then calculates the change in ULC around the stations. Using this methodology, we estimated the change in land cover in a 100 m radius around 48 Swedish meteorological stations during the period 1960-2019. The seasonal Mann-Kendall test, together with the Theil-Sen estimator and linear regression were applied on the stations’ long-term temperature and precipitation data in order to determine systematic differences in climatic trends between stations with varying degrees of ULC increase, and to explore the possible influence of urban heat islands. Initial results associate large increases in ULC with higher positive Theil-Sen estimator values for temperature observations, and negative linear regression slopes in precipitation observations, respectively (p < 0.001). Thus, the temperature increase has been more pronounced at meteorological stations experiencing substantial ULC increase. Conversely, these stations showed decreasing trends in precipitation. Overall, our results show a correlation between the change in ULC around climatic stations and their long-term trends in climatic observations, and suggest possible influences of urban heat island effects on observed climate data in Sweden.