Spatio-Temporal Distribution And Drivers Of Urban Heat Island Intensity And Footprint In Beijing-Tianjin-Hebei Urban Agglomeration

The rise of land surface temperature(LST)and population caused by urbanization has led to a series of frequent events such as urban heat island(UHI)effect,extreme climate and heat waves,which threaten human health seriously.Compared with the traditional MODIS LST data,the 1 km daily all-weather land surface temperature dataset(TRIMS LST)used in this study has a more accurate precision.Based on this dataset,this thesis analyzed the spatial and temporal distribution characteristics of surface urban heat island intensity(SUHII)and surface urban heat island footprint(SUHIF)in the Beijing-Tianjin-Hebei urban agglomeration by using quantitative remote sensing,geospatial analysis,and geo-statistical technologies,combined with the urban boundary dataset.Meanwhile,based on DEM,nightlight,statistical data,Enhance Vegetation Index(EVI),and PM2.5,combined with multiple linear regression,random forest regression,and principal component analysis methods,the drivers of SUHII and SUHIF were discussed.The main conclusions are as follows:(1)Based on the TRIMS LST data set and Arc GIS 10.4 software,the spatial and temporal distribution map of LST in the Beijing-Tianjin-Hebei urban agglomeration was cut out.It is found that the diurnal LST decreases from north to south during 2005-2018,which is highly consistent with DEM spatial distribution and presents an opposite trend.That is,LST is low where DEM is high,and LST is high where DEM is low.In addition,the annual mean LST in the daytime(>10℃)was higher than that at night(<10℃).In terms of time,LST during 2005-2018 showed a downward trend and then an upward trend,and the turning point was 2010,which was mainly due to the continuous large-scale invasion of Arctic cold air to the south since the winter of 2009,which accumulated in the middle and high latitudes of the northern Hemisphere and caused frequent cooling in North China.However,the temperature rise after 2010 was mainly due to the accelerated urbanization process and the continuous increase of urban construction land area(from 2,695 km~2 in 2005 to 3,991 km~2 in 2018),which resulted in the continuous urban temperature rise.(2)As for the heat island intensity,the SUHII in the daytime was higher than that at night.In 2005,2010,2015,and 2018,mean daytime SUHII was 0.21℃,0.03℃0.35℃,and0.53℃higher than that at night,respectively.The SUHII values in Beijing were the highest and all were greater than 3℃,and it also showed an increasing trend from 2005 to 2018,which was closely related to the rapid development of Beijing in the past 20 years.Other cities showed the lowest SUHII in 2010 and the second in 2018,mainly because the cold air in the Arctic moved south in 2009 and a series of measures of ecological civilization construction in 2018 weakened SUHII to some extent.(3)As for Surface urban heat island footprint(SUHIF),the maximum SUHIF during daytime from 2005 to 2018 was mainly concentrated in densely populated areas such as Beijing,Tianjin,and Shijiazhuang,all of which were larger than 7 km.The minimum value was concentrated in Handan,Hengshui,and Langfang cities,all less than 3 km,although Hengshui and Langfang cities’population density is larger,the area of construction land administrative scope is small,which is the main reason for their small SUHIF.At night,the maximum SUHIF of Tianjin,Baoding,Tianjin,and Beijing cities from 2005 to 2018 were also greater than 7 km,and the minimum SUHIF of Hengshui,Zhangjiakou,Hengshui,and Qinhuangdao cities was also less than 3 km,which was consistent with the reason of daytime SUHIF and is affected by population density,construction land area,and other factors.(4)Based on the principal component analysis method,the influencing factors of SUHII and SUHIF were analyzed.The results showed that in 2005,2010,2015,and 2018,the first three principal components contained more than 85%of the information of the original variables.Therefore,in this thesis,the first three principal components were selected to calculate the table of load coefficient of each factor during daytime and nighttime from 2005to 2018.The results showed that the representative variables of the first three principal components were Population Density(PD),PM2.5,and Enhance Vegetation Index(EVI),respectively.Combined with the random forest regression method,the contribution of PM2.5to SUHII in the daytime and PD at night was the highest expected in 2010.For SUHIF,the contribution rate of EVI,PM2.5,and PD in daytime showed different results in different years,but all of them were greater than 20%,while at night,PD made the greatest contribution to SUHIF except in 2010.In addition,the results of multiple linear regression showed that for SUHII,PM2.5 was negatively correlated with SUHII during the daytime from 2005 to 2018,while PD and EVI were positively correlated with SUHII,and the correlation coefficient showed a trend of PM2.5>PD>EVI.For SUHIF,at night in 2015,the coefficients of PM2.5,PD,and EVI were-0.523,0.875,and-0.452,respectively,indicating that PM2.5 and EVI were negatively correlated with SUHIF,while PD and SUHIF were positively correlated.