Using Prophet Forecasting Model to Characterize the Temporal Variations of Historical and Future Surface Urban Heat Island in China

摘要

The changes in surface urban heat island (SUHI) are usually diurnal variations, seasonal variations, and interannual trends, yet previous temporal analysis methods have difficulties in decomposing time-series SUHI intensity (SUHII) into multiple time scales. This study proposed a prophet forecasting model to decompose time-series SUHII in 102 Chinese cities for the analysis of temporal variations. The occurrence time of SUHII extremums and annual SUHII amplitude (ASA) were investigated. Results show that the maximum daytime SUHII in 88 of the cities occurred in summer, and the minimum daytime SUHII in 80 of the cities occurred in winter. At night, the maximum SUHII in northern cities occurred in the cold season, while the minimum SUHII occurred in the hot season, which is the opposite of the results of southern cities. Larger ASA was observed in northern cities than southern cities,and the ASA was larger during the day than night. The interannual trend of SUHII was characterized and the SUHII in 2030 was predicted. Results show that the increasing trends of SUHII were found in 78 and 83 of the cities in the day and night, respectively. The interannual trend of daytime SUHII was negative (-0.01 ± 0.28°C/10 yr) in northern cities, while it was positive (0.22 ± 0.20°C/10 yr) in southern cities. At night, northern cities (0.19 ± 0.12°C/10 yr) had a faster increasing trend of SUHII than southern cities (0.06 ± 0.13°C/10 yr). The SUHII in China will be more evident in 2030, and various mitigation measures are needed for SUHII mitigation.