Impact of seasonal variations in weekday electricity use on urban air temperature observed in Osaka, Japan

Seasonal variations in the human activity signal (HAS) affecting the urban temperature were analysed in districts with offices and commercial buildings, apartment buildings, and single-family households in the urban area of Osaka, Japan. Temperatures were observed at building rooftops and at ground level in 14 districts. Spatial and temporal patterns of HAS were identified for a combined spring and autumn intermediate season, the summer cooling season, and the winter heating season by observing the temperature differences between weekdays and weekends after discounting the naturally induced temperature signal. A positive HAS (higher weekday temperatures) was observed during daytime hours in summer and the values around 1200h JST (UTC + 9 h) exceeded 1 degrees C at the downtown office and mixed-building districts and 0.5-0.7 degrees C in the surrounding inland residential districts. However, a negative HAS (lower weekday temperatures) extended throughout all districts during daytime hours in winter. The maximum negative values ranged from -0.3 to -0.4 degrees C between 1200 and 1500h. Statistical analyses of 30-year HAS data provided sufficient evidence that an increase in heat pump air-conditioning caused negative HAS in winter daytime hours as well as positive HAS in summer daytime hours. While HAS was not observed at night-time in all districts during summer, positive HAS in winter had values of 0.3-0.4 degrees C between 0000 and 0200h and was especially prominent at inland residential districts. This night-time positive HAS in the heating season was most likely due to a gradual ventilation of heated indoor air and an outgoing sensible heat transfer from the walls of buildings where heating systems were operated. Seasonal variation of HAS was related to the energy consumption due to air-conditioning, and the expected temperature increase due to the waste heat was predicted using a simple one-dimensional mixing-layer model.