Spatial and temporal variations of the SRHII at daytime and nighttime
Significant seasonal differences are observed in the SRHII in the YRDUA (Figure A1 and A2, Appendix A). In the daytime, RHI was concentrated in the Nanjing, “Su-Xi-Chang”, Ningbo, Shanghai, and Hangzhou metropolitan areas. Due to the high built-up areas and PD, the distribution of surface RHI is denser and stronger than that in the north and southwest of the YRDUA. The built-up area can absorb heat and store heat energy, which makes the surface temperatures rise rapidly. In spring and autumn, the spatial distribution of the RHI in spring or autumn was similar to that in summer except the spatial extent was tapered. However, the RHI gradually shrinks and transfers to the southern area of the YRDUA in winter, such as Linhai and Ningbo City, which is due to the relatively high solar radiation of the geographic location of the southern cities. The distance of the RHI is gradually shortened between cities and even into one piece from 2003 to 2017 due to long-term urban expansion and rapid growth of construction land (Figure A1, Appendix A). In the nighttime, the spatial pattern of the RHI is very different from that of the daytime. RHI mainly concentrates on Taihu Lake, Dianshan Lake, Ge Lake in the center part, Hongzhe Lake in the northwest, and Qiandao Lake in the southwest. Because water has a high specific heat capacity, it has the function of preserving heat at nighttime. Some cities like Shanghai, Hangzhou, and Nanjing have the strongest heat island in winter and the weakest heat island in summer. Urban areas usually have dense buildings, PD, and energy emissions, so there are more energy emissions at night. High surface albedo in urban areas at night leads to lower heat storage4,40 and ultimately resulting in smaller UHI at nighttime (Figure A2, Appendix A).
From spring to summer and then summer to winter, RHI increases first and then decreases, and it reaches a peak in summer. For example, the proportion of the RHI was 12.65%, 31.03%, 21.12%, and 5.49% in spring, summer, autumn, and winter in 2017, respectively (Fig. 2d). An upward trend in the area of the RHI is observed from 2003 to 2017 in summer. In detail, the proportion of the heat island zone is 21.74%, 22.17%, and 31.03% in the summer of 2003, 2010, and 2017, respectively (Fig. 2d). It is because the urban areas of YRDUA have increased from 3571.01 km2 to 8760.26 km2 in 2003 and 2017, respectively (Figure B1, Appendix B). Moreover, the area of the medium heat island and strong heat island increased by 41.08% and 66.40% from 2003 to 2017 (Fig. 2b,c). A gradual decreasing trend is observed for the four grades of the SRHII (2–4 °C, 4–6 °C, > 6 °C, > 2 °C) in winter from 2003 to 2017 (Fig. 2a–d). The area of the RHI in winter was 18,481 km2, 8640 km2, and 6280 km2 in 2003, 2010, and 2017, respectively (Fig. 2d). Vegetation coverage is low in winter and bare soil is formed after harvest. It leads to the RHI decrease in winter. The above results indicated that the SRHII became increasingly hot in summer and increasingly cold in winter and that the trend became more obvious as the SRHII increased in the ranges of 2–4 °C, 4–6 °C, > 6 °C. However, the seasonal variation of the RHI in the nighttime is opposite to that in the daytime. From spring to summer and then to winter, the area of the RHI decreases first and then increases, and it falls in the lowest value in summer (Fig. 2e–g). For example, the area of RHI is 19,209 km2, 5659 km2, 34,621 km2, and 38,596 km2 in spring, summer, autumn, and winter in 2017, respectively (Fig. 2h). The annual average of RHI regular increases, with values of 17,510 km2, 20,042 km2, and 20,097 km2 in 2003, 2010, and 2017, respectively (Fig. 2h).
Seasonal and inter-annual variations of the SRHII during the daytime (a–d) and nighttime (e–h) of the YRDUA.
Relationship between the SRHII and influencing factors
Results showed surface biophysical factors have a higher correlation with SRHII than socio-economic factors and climate factors in the day and night. NDBI and EVI have a stronger effect on SRHII than other biophysical factors in the day. NDBI showed a significant positive correlation with SRHII, while EVI showed a negative correlation with SRHII. In detail, NDBI (r = 0.567, p < 0.001) was the highest correlation with RHI in summer, followed by EVI (r = − 0.54, p < 0.001) (Fig. 3a). The correlation of EVI was highest in spring (r = − 0.577, p < 0.001). NDBI also showed the highest correlation in autumn (r = 0.425, p < 0.001) (Fig. 3c). White-sky Albedo (WSA) has a positive correlation in each season. MNDWI has a positive effect on the cooling heat environment in the day for its high specific heat capacity. However, the correlation of biophysical factors in the day was different at night. MNDWI has the highest correlation (positive correlation) with SRHII, followed by the EVI (negative correlation) at night. The correlation between the MNDWI and SRHII in spring, summer, autumn, and winter was 0.844, 0.558, 0.725, and 0.492 (Fig. 3f) respectively. NDBI has a low degree of interpretation of SRHII at night compared to other biophysical factors, indicating that the effect of building density on SRHII at night was weak. Zhou11 also showed that a weak negative correlation between UHI intensity and built-up intensity was observed in the YRDUA at night. WSA was negatively associated with SRHII at night. Compared with the conclusions in the daytime show that during the day, water bodies have a cooling effect on RHI in the day and a warming effect at night while the trend of WSA was the opposite. Vegetation has the effect of relieving the thermal environment around the clock. It indicated that water bodies were not the best choice for relieving the thermal environment. In addition, As SRHII increases from weak (2 °C < SRHII ≤ 4 °C), medium (4 °C < SRHII ≤ 6 °C) to strong (SRHII > 6 °C), the correlation coefficient (r) of WSA and NDBI increases while EVI and MNDWI have opposite trends. It indicated that controlling building density and changing the building materials, increasing the proportion of vegetation and water bodies can mitigate RHI, especially in areas with high and extremely high temperatures.
Pearson correlation coefficients of different grades of the SRHII (2–4 °C, 4–6 °C, > 6 °C, > 2 °C) during the daytime (a,c,e) and nighttime (b,d,f) of 2003, 2010 and 2017.
In the socio-economic factors layers, anthropogenic heat emission can’t be ignored in the spatial distribution of RHI in the day. It has the highest correlation with RHI in spring, summer, and autumn, with 0.534, 0.582, and 0.366, respectively. PD has a positive correlation to RHI with (r = 0.278, p < 0.001) (Fig. 3a). Per capita gross domestic product (PCGDP) has no significant correlation with the RHI in all seasons, indicating that the influence of economic development level on the change of RHI is weak. At night, anthropogenic heat emissions in spring and summer were also the most relevant to SRHII, at 0.28 and 0.295 (Fig. 3f), respectively. The correlation between PCGDP and SRHII at night was highest in autumn and winter, at 0.183 and 0.178, respectively. PD has a weak effect on SRHII at night. With the increase of SRHII from weak, medium to strong, the correlation increased between anthropogenic heat emission and SRHII, which means that the effect of artificial heat discharge becomes a more important role with the increase of SRHII.
In the climate factors layers, air temperature is positively correlated with SRHII while PRE is negatively correlated with SRHII during the day. For winter, these factors are closely related to SRHII distribution than biophysical and socio-economic factors. Specifically, AT (r = 0.343, p < 0.001) became the highest correlated factor with SRHII in winter, followed by PRE (r = 0.234, p < 0.001) (Fig. 3a). The correlation of AT and PRE and SRHII was weak at night in each season (Fig. 3f).
In summary, NDBI, EVI, and anthropogenic heat emission have related closely to RHI and present seasonal variations in the day, with summer > spring > autumn. Climate (AT and PRE) factors can explain more spatial differences of RHI in winter. Water bodies and surface albedo have a negative and positive effect on the SRHII in the daytime and nighttime. With the increase of SRHII from weak, medium to strong, the correlation coefficient (r) of SRHII and albedo, NDBI, and anthropogenic heat emission increases while EVI and MNDWI have the opposite trends.
Changes in dominant factors of the SRHII
The results of the stepwise regression show that the EVI and NDBI have the largest independent contribution relative to the other factors in the variation of SRHII except in winter. The most important independent variable factor affecting the SRHII is the EVI in the spring daytime of 2003, and it had an explanatory rate of 33.3% (Fig. 4a). However, the dominant explanatory factor changed to the NDBI (explanatory rate is 20.7) in the spring daytime of 2010 and the EVI (explanatory rate is 20.1%) in the spring daytime of 2017. The total explanation rates increased as the SRHII increased. For example, the total explanatory rates affecting the SRHII in 2010 spring daytime are 9.7%, 11.1%, and 14.7% in the range of 2–4 °C, 4–6 °C and > 6 °C, respectively. NL has the largest interpreted independent contribution (33.8%) on the summer daytime of 2003, although the dominant influencing factor shifted to the NDBI (37.3% and 31.7%) in 2010 and 2017. In the summer daytime of 2017, the total explanatory rates were 5.3%, 6.7%, and 12.3% for SRHII in the range of 2–4 °C, 4–6 °C and > 6 °C, respectively (Fig. 4c).
Explanation degree of the stepwise linear regression analysis of the SRHII (°C) and impact factors during the daytime (a,c,e,g) and nighttime (b,d,f,h) in spring (a,b), summer (c,d), autumn (e,f), and winter (g,h).
For autumn daytime, the most important independent factor affecting the SRHII is the NDBI, whose explanatory rate was 18% in 2003. The dominant explanatory variable was also the NDBI in 2010, and its explanatory rate was 26.8%. However, the largest explanatory factor shifted to the EVI (explanatory rate was 22.8%) in 2017 (Fig. 4e). In winter, the total explanatory rate fluctuated in 2003, 2010, and 2017. Compared with other seasons in the daytime, PRE had a large explanatory rate (26.2%) for the SRHII in the winter of 2003. In 2010, the main explanatory variable was AT (5.4%) (Fig. 4g).
At night, MNDWI was the dominant explanatory variable in spring, autumn and winter, while EVI was the main explanatory variable in summer. In detail, MNDWI showed the largest interpreted independent contribution to the SRHII in spring, and its explanatory rate was 54.1%, 57%, and 71.2% in 2003, 2010, and 2017, respectively (Fig. 4b). In the summer, the dominant explanatory parameter was the EVI in 2003, 2010, and 2017, and it explained 37.4%, 46.5%, and 42%, respectively (Fig. 4d). MNDWI also had the largest interpreted independent contribution in 2003, 2010, and 2017, and it explained 33.7%, 51%, and 52.6%, respectively in autumn (Fig. 4f). In winter, the MNDWI had the largest interpreted independent contribution in 2003, 2010, and 2017, and its explanatory rate was 21.5%, 11.2%, and 24.2%, respectively (Fig. 4h). The total explanatory rates also increased as the SRHII increased from weak, medium to strong in each season.
In summary, NDBI, EVI, and anthropogenic heat emission were the key parameters for determining the daytime RHI when the relative temperature was high in YRDUA, while AT and PRE were the dominant explanatory variable in winter. The total explanatory rates affecting the daytime SRHII present seasonal variations, with summer > spring > autumn > winter. The dominant influencing factor was the MNDWI in spring, autumn and winter, while EVI had the largest contribution in summer at night.
Source: Ecology - nature.com
