Research on the relationship between public open space and the thermal environment in Shanghai based on multisource data integration

Rapid urbanization has increased the severity of the urban thermal environment, and public open space (POS) plays an important role in improving the urban thermal environment. Previous studies focused on the cooling effect of a single type of POS but seldom paid attention to the comprehensive cooling/warming effect and heterogeneity of multi-type POSs. This paper combines multisource data to classify and extract the POSs in Shanghai. Through remote sensing image inversion of the urban surface temperature, the spatial and temporal evolution of urban heat islands (UHIs) in Shanghai was analysed. By constructing an evaluation index system, the correlations among the internal thermal environment, external thermal environment and influencing factors of POS are revealed. The results indicate that (1) the UHI effect in Shanghai is no longer limited to the city centre but extends to the peripheral areas, with obvious characteristics of "inner bulge and outer depression" and "south cold and north hot, east cold and west hot". (2) Urban parks can improve the thermal environment to a certain extent, but sports grounds and square land can cause small-scale UHIs. A total of 99.06 % of urban parks have cooling effects, the average cooling distance is 199.33 m, and the average cooling intensity is 2.83 °C. A total of 86.67 % of the sports venues have a warming effect, and the influence distance (207.15 m) is slightly greater than that of urban parks. (3) The average value of the enhanced impervious surface index (ENDISI) in urban parks has a significant positive effect on land surface temperature (LST), whereas the circumference and water body proportion have a significant negative effect on LST. The average LST of square land is negatively correlated with the proportion of forest area and positively correlated with the ENDISI. There was a significant positive correlation between the average LST inside the sports venue and its external ENDISI. When the external ENDISI decreases by 0.0423, the average LST inside the sports venue decreases by 1 °C. (4) For cold island spaces with cooling effects, the greater the proportions of forestland within the samples, the greater the mean values of the normalized difference vegetation index (NDVI), the smaller the proportions of impervious surface, and the smaller the mean values of the ENDISI are, the greater the cooling distances to the external environment. For every 1.36 °C decrease in the LST within the sample, every 0.29 increase in the mean NDVI, every 0.09 decrease in the mean normalized difference water index (NDWI) of the external environment, and every 0.04 increase in the mean ENDISI of the external environment, the cooling amplitude of the sample to the external thermal environment increases by 1 °C. The research results provide data that can be used to improve the urban thermal environment and cope with the challenges of climate change.