基于遥感与机器学习的中国城市夏季热岛气候带分异特征（2003-2022年）及其驱动因子分析

doi:10.16258/j.cnki.1674-5906.2025.10.011

生态环境学报 ›› 2025, Vol. 34 ›› Issue (10): 1609-1617.DOI: 10.16258/j.cnki.1674-5906.2025.10.011

• 研究论文【环境科学】 • 上一篇下一篇

基于遥感与机器学习的中国城市夏季热岛气候带分异特征（2003-2022年）及其驱动因子分析

王恭斌¹^,²(), 花利忠¹^,²^,^*(), 卢璇¹^,², 李琳¹^,², 章欣欣¹, 李兰晖¹^,²

1.厦门理工学院计算机与信息工程学院，福建厦门 361024
2.厦门理工学院空间信息技术研究所，福建厦门 361024

收稿日期:2025-03-24 出版日期:2025-10-18 发布日期:2025-09-26
通讯作者: *E-mail: lzhua@xmut.edu.cn
作者简介:王恭斌（1997年生），男，硕士研究生，主要从事城市热环境研究。E-mail: 13023909076@163.com
基金资助:
福建省自然科学基金项目(2020J01265);福建省自然科学基金项目(2023J011429)

Spatiotemporal Patterns and Drivers of Summer Urban Heat Islands in Chinese Cities (2003-2022): A Multi-Climate Zone Analysis Using Remote Sensing and Machine Learning

WANG Gongbin¹^,²(), HUA Lizhong¹^,²^,^*(), LU Xuan¹^,², LI Lin¹^,², ZHANG Xinxin¹, LI Lanhui¹^,²

1. School of Computer and Information Engineering, Xiamen University of Technology, Xiamen 361024, P. R. China
2. Spatial Information Technology Institute of Xiamen University of Technology, Xiamen 361024, P. R. China

Received:2025-03-24 Online:2025-10-18 Published:2025-09-26

摘要/Abstract

摘要：

在全球变暖与快速城市化协同作用下，揭示城市热岛效应（UHI）的时空异质性及其气候带依赖性机制对准确制定缓解策略至关重要。然而，当前研究对UHI驱动因子的气候带特异性及昼夜耦合效应仍缺乏系统性认识。基于2003-2022年MODIS遥感数据、气象观测及社会经济数据，运用随机森林模型与Sen+MK检验趋势分析，系统评估柯本气候分类下暖温带、冷温带和干带的中国65个城市夏季地表城市热岛强度（SUHII）时空分异规律及驱动机制。研究发现，1）空间格局：昼夜SUHII呈现显著气候带分异，呈“昼降夜升”梯度反转规律，即昼间SUHII随气候区干旱性增强递减（暖温带4.98 ℃>冷温带4.30 ℃>干带2.16 ℃），夜间呈反向分布（干带2.92 ℃>冷温带2.29 ℃>暖温带1.83 ℃）。2）时间演变：54%的城市昼间SUHII呈下降趋势（显著区域集中于暖温带），而48%的城市夜间SUHII显著上升的区域主要位于冷温带南部和暖温带北部。3）驱动机制：城乡白空反照率差异（ΔWSA）为全域主导因子，但气候带分异显著——暖温带昼夜受ΔWSA、降水（PRE）与高程（DEM）共同调控，冷温带昼间响应植被差异（ΔEVI）而夜间受人口（POP）驱动，干带昼夜均以蒸散发（ET）为主导。研究从气候带视角揭示中国SHUII昼夜分异的梯度规律与驱动机制异质性，为制定气候带适配的热岛缓解方案提供科学依据。

关键词: 城市热岛效应, 时空特征, 驱动因子, 遥感数据, 随机森林, 气候带

Abstract:

Understanding the spatiotemporal heterogeneity and climate zone-dependent mechanisms of urban heat island (UHI) effects is critical for developing targeted mitigation strategies under global warming and rapid urbanization. However, existing studies lack comprehensive analyses of climate-specific variations and diurnal and nocturnal interactions of UHI driving factors. Thus, this study utilized a random forest model combined with Sen's slope estimator and Mann-Kendall Test (Sen+MK) trend analysis to systematically investigate the spatiotemporal patterns and driving mechanisms of summer surface urban heat island intensity (SUHII) across 65 Chinese cities spanning temperate, cold, and arid Köppen climate zones. Using MODIS remote sensing data, meteorological observations, and socioeconomic indicators from 2003 to 2022, we revealed three key findings: 1) Spatial Patterns: The daytime and nighttime SUHII exhibited distinct differentiation across climate zones, that is, a “day decrease-night increase” reversal pattern. Specifically, SUHII decreased with increasing aridity of climate zones during the daytime (temperate zone 4.98 ℃>cold zone 4.30 ℃>arid zone 2.16 ℃) and at night (arid zone 2.92 ℃> cold zone 2.29 ℃>temperate zone 1.83 ℃). 2) Temporal Trends: A decreasing trend was observed in 54% of cities in daytime SUHII (p<0.01, with significant areas concentrated in the temperate zone), while 48% of cities exhibited a significant increase in nighttime SUHII, mainly in the southern cold and northern temperate zones. 3) Driving Mechanisms: The urban-rural white sky albedo difference (ΔWSA) emerged as the dominant universal driver, although significant climate zone-specific variations were observed. Urban greening and land-use strategies should prioritize ΔWSA mitigation through surface material optimization, green coverage expansion, and impervious surface reduction to regulate surface heat accumulation and mitigate the effects of UHI. In temperate zones, the SUHII is diurnally regulated by ΔWSA, precipitation (PRE), and elevation (DEM). Constructed wetlands and integrated water management can enhance evapotranspiration and effectively moderate temperature variations. Cold temperate zones exhibited distinct patterns: daytime SUHII responded to vegetation differences (ΔEVI), whereas nighttime intensity was driven by population density (POP). This requires targeted mitigation measures, including urban core population control, functional decentralization, and high-albedo facade retrofits. Arid zones are consistently dominated by evapotranspiration (ET) during the diurnal cycle. These findings provide a climate-zone-differentiated perspective on SUHII dynamics, enabling the development of regionally tailored urban heat mitigation strategies.

Key words: surface urban heat island (SUHII), spatiotemporal patterns, driving factors, remote sensing data, random forest, climate zones

中图分类号:

王恭斌, 花利忠, 卢璇, 李琳, 章欣欣, 李兰晖. 基于遥感与机器学习的中国城市夏季热岛气候带分异特征（2003-2022年）及其驱动因子分析[J]. 生态环境学报, 2025, 34(10): 1609-1617.

WANG Gongbin, HUA Lizhong, LU Xuan, LI Lin, ZHANG Xinxin, LI Lanhui. Spatiotemporal Patterns and Drivers of Summer Urban Heat Islands in Chinese Cities (2003-2022): A Multi-Climate Zone Analysis Using Remote Sensing and Machine Learning[J]. Ecology and Environmental Sciences, 2025, 34(10): 1609-1617.

图/表 11

参考文献 30

[1]	BECK H E, MCVICAR T R, VERGOPOLAN N, et al., 2023. High-resolution (1 km) Köppen-Geiger maps for 1901-2099 based on constrained CMIP6 projections[J]. Scientific Data, 10(1): 724.
[2]	DOS SANTOS A R, DE OLIVEIRA F S, DA SILVA A G, et al., 2017. Spatial and temporal distribution of urban heat islands[J]. Science of the Total Environment, 605-606: 946-956.
[3]	GENG X L, ZHANG D, LI C W, et al., 2023. Impacts of climatic zones on urban heat island: Spatiotemporal variations, trends, and drivers in China from 2001-2020[J]. Sustainable Cities and Society, 89: 104303.
[4]	HAN D L, ZHANG T T, QIN Y X, et al., 2022. A comparative review on the mitigation strategies of urban heat island (UHI): A pathway for sustainable urban development[J]. Climate and Development, 15(5): 379-403.
[5]	KIM H, GU D, KIM H Y, 2018. Effects of urban heat island mitigation in various climate zones in the United States[J]. Sustainable Cities and Society, 41: 841-852.
[6]	LI H D, ZHOU Y Y, LI X M, et al., 2018. A new method to quantify surface urban heat island intensity[J]. Science of The Total Environment, 624: 262-272.
[7]	LI K N, CHEN Y H, JIANG J B, 2023. Grading surface urban heat island and investigating factor weight based on interpretable deep learning model across global cities[J]. Environment International., 180: 108196.
[8]	LI L, ZHA Y, ZHANG J H, 2020. Spatially non-stationary effect of underlying driving factors on surface urban heat islands in global major cities[J]. International Journal of Applied Earth Observation and Geoinformation, 90: 102131.
[9]	LI X M, ZHOU Y Y, ASRAR G R, et al., 2017. The surface urban heat island response to urban expansion: A panel analysis for the conterminous United States[J]. Science of the Total Environment, 605-606: 426-435.
[10]	LIU F, LIU J, ZHANG Y Q, et al., 2024. Construction of a cold island network for the urban heat island effect mitigation[J]. Science of the Total Environment, 915: 169950.
[11]	OKE T R, 1982. The energetic basis of the urban heat island[J]. Quarterly Journal of the Royal Meteorological Society, 108(455): 1-24.
[12]	O'MALLEY C, KIKUMOTO H, 2022. An investigation into heat storage by adopting local climate zones and nocturnal-diurnal urban heat island differences in the Tokyo Prefecture[J]. Sustainable Cities and Society, 83: 103959.
[13]	PENG X X, ZHOU Y Y, FU X H, et al., 2022. Study on the spatial-temporal pattern and evolution of surface urban heat island in 180 shrinking cities in China[J]. Sustainable Cities and Society, 84: 104018.
[14]	RINCHUMPHU D, SRIVANIT M, IAMCHUEN N, et al., 2024. Exploring summer variations of driving factors affecting land use zoning based on the surface urban heat island in Chiang Mai, Thailand[J]. ISPRS International Journal of Geo-Information, 13(7): 228.
[15]	SHE Y H, LIU Z H, ZHAN W F, et al., 2022. Strong regulation of daily variations in nighttime surface urban heat islands by meteorological variables across global cities[J]. Environmental Research Letters, 17(1): 14049.
[16]	WAN Y, DU H, YUAN L, et al., 2025. Exploring the influence of block environmental characteristics on land surface temperature and its spatial heterogeneity for a high-density city[J]. Sustainable Cities and Society, 118: 105973.
[17]	WANG L Q, DE BOECK H J, CHEN L X, et al., 2022. Urban warming increases the temperature sensitivity of spring vegetation phenology at 292 cities across China[J]. Science of the Total Environment, 834: 155154.
[18]	XIE P Y, WANG H, 2021. Potential benefit of photovoltaic pavement for mitigation of urban heat island effect[J]. Applied Thermal Engineering, 191: 116883.
[19]	YANG J, HUANG X, 2021. The 30 m annual land cover dataset and its dynamics in China from 1990 to 2019[J]. Earth System Science Data, 13(8): 3907-3925.
[20]	YAO R, WANG L C, HUANG X, et al., 2019. Greening in rural areas increases the surface urban heat island intensity[J]. Geophysical Research Letters, 46(4): 2204-2212.
[21]	ZHOU D C, ZHAO S Q, LIU S G, et al., 2014. Surface urban heat island in China’s 32 major cities: Spatial patterns and drivers[J]. Remote Sensing of Environment, 152: 51-61.
[22]	曹畅, 李旭辉, 张弥, 等, 2017. 中国城市热岛时空特征及其影响因子的分析[J]. 环境科学, 38(10): 3987-3997.
	CAO C, LI X H, ZHANG M, et al., 2017. Analysis of the spatiotemporal characteristics and influencing factors of urban heat islands in China[J]. Environmental Science, 38(10): 3987-3997.
[23]	侯赛男, 党国锋, 2022. 机器学习模拟热环境及热岛时空变化特征研究[J]. 测绘科学, 47(9): 200-207.
	HOU S N, DANG G F, 2022. Research on machine learning simulation of thermal environment and spatiotemporal variation characteristics of heat islands[J]. Science of Surveying and Mapping, 47(9): 200-207.
[24]	贾文茜, 任国玉, 于秀晶, 等, 2021. 中国东部季风区不同气候带城市热岛效应的差异[J]. 气候与环境研究, 26(5): 569-582.
	JIA W Q, REN G Y, YU X J, et al., 2021. Differences in urban heat island effects among different climate zones in the eastern monsoon region of China[J]. Climatic and Environmental Research, 26(5): 569-582.
[25]	刘宇翔, 杨英宝, 胡佳, 等, 2022. 基于长时序MODIS数据的中国城市昼夜热岛强度时空特征[J]. 地球信息科学学报, 24(5): 981-995. DOI
	LIU Y X, YANG Y B, HU J, et al., 2022. Spatiotemporal characteristics of daytime and nighttime urban heat island intensity in China based on long-term MODIS data[J]. Journal of Geo-information Science, 24(5): 981-995.
[26]	孙艳伟, 王润, 郭青海, 等, 2021. 基于人居尺度的中国城市热岛强度时空变化及其驱动因子解析[J]. 环境科学, 42(1): 501-512.
	SUN Y W, WANG R, GUO Q H, et al., 2021. Analysis of spatiotemporal variation and driving factors of urban heat island intensity in China based on human settlement scale[J]. Environmental Science, 42(1): 501-512.
[27]	孙喆, 2020. 高密度城区形态要素对热环境的影响作用——以北京市五环内区域为例[J]. 生态环境学报, 29(10): 2020-2027. DOI
	SUN Z, 2020. Impact of urban morphology factors on thermal environment in high density urban areas: A case of Beijing within 5th ring road[J]. Journal of Ecology and Environment, 29(10): 2020-2027.
[28]	王嘉丽, 冯婧珂, 杨元征, 等, 2023. 南宁市主城区不透水面与热环境效应的空间关系研究[J]. 生态环境学报, 32(3): 525-534. DOI
	WANG J L, FENG J K, YANG Y Z, et al., 2023. Research on spatial relations between impervious surfaces and the urban thermal environment in the central metropolitan area of Nanning city[J]. Journal of Ecology and Environment, 32(3): 525-534.
[29]	王美雅, 徐涵秋, 2018. 中国大城市的城市组成对城市热岛强度的影响研究[J]. 地球信息科学学报, 20(12): 1787-1798. DOI
	WANG M Y, XU H Q, 2018. Research on the impact of urban composition on urban heat island intensity in large cities in China[J]. Journal of Geo-information Science, 20(12): 1787-1798.
[30]	王阳, 孙然好, 2021. 区域气候背景对城市热岛效应的影响规律[J]. 生态学报, 41(11): 4288-4299.
	WANG Y, SUN R H, 2021. The Influence of regional climate background on urban heat island effects[J]. Acta Ecologica Sinica, 41(11): 4288-4299.

名称	类型	空间分辨率	来源	时间范围
城市边界	矢量	-	自然资源部标准地图	2020
土地利用数据集	栅格	30 m	https://zenodo.org/	2003-2022
地表温度数据	栅格	1 km	MYD11A2	2003-2022
增强植被指数	栅格	1 km	MOD13A2	2003-2022
蒸散发	栅格	1 km	MOD16A2	2003-2022
夜间灯光指数	栅格	1 km	DMSP/OLS	2003-2012
夜间灯光指数	栅格	500 m	NPP/VIIRS	2013-2022
高程	栅格	90 m	SRTM
人口数据集	栅格	1 km	国家统计年鉴	2003-2022
气溶胶光学厚度	栅格	1 km	MCD19A2_ GRANULES	2003-2022
降水量	栅格	5.6 km	UCSB-CHG	2003-2022
白空反照率	栅格	500 m	MCD43A3	2003-2022

名称	类型	空间分辨率	来源	时间范围
城市边界	矢量	-	自然资源部标准地图	2020
土地利用数据集	栅格	30 m	https://zenodo.org/	2003-2022
地表温度数据	栅格	1 km	MYD11A2	2003-2022
增强植被指数	栅格	1 km	MOD13A2	2003-2022
蒸散发	栅格	1 km	MOD16A2	2003-2022
夜间灯光指数	栅格	1 km	DMSP/OLS	2003-2012
夜间灯光指数	栅格	500 m	NPP/VIIRS	2013-2022
高程	栅格	90 m	SRTM
人口数据集	栅格	1 km	国家统计年鉴	2003-2022
气溶胶光学厚度	栅格	1 km	MCD19A2_ GRANULES	2003-2022
降水量	栅格	5.6 km	UCSB-CHG	2003-2022
白空反照率	栅格	500 m	MCD43A3	2003-2022

气候带	白天		夜晚
气候带	R²	RMSE	R²	RMSE
暖温带（n=32）	0.91	0.40	0.90	0.22
冷温带（n=26）	0.91	0.41	0.88	0.14
干带（n=7）	0.89	0.56	0.73	0.19
均值	0.90	0.46	0.84	0.18

气候带	白天		夜晚
气候带	R²	RMSE	R²	RMSE
暖温带（n=32）	0.91	0.40	0.90	0.22
冷温带（n=26）	0.91	0.41	0.88	0.14
干带（n=7）	0.89	0.56	0.73	0.19
均值	0.90	0.46	0.84	0.18

驱动因子	白天			夜晚
驱动因子	暖温带	冷温带	干带	暖温带	冷温带	干带
POP	−0.28	−0.10	0.16	0.03	0.65**	−0.23
△EVI	−0.45**	−0.76**	−0.54**	−0.57*	0.28	0.43
NL	0.03	−0.12	0.47	−0.38	0.19	−0.25
DEM	−0.12	−0.33	0	−0.46**	0.33	0.22
AOD	−0.04	−0.13	−0.44	−0.06	−0.35	−0.04
ET	0.56**	0.76**	0.11	0.06	0.08	−0.37
PRE	0.46**	0.17	0.43	0.43*	0.39	−0.32
△WSA	−0.49**	−0.58**	−0.54**	−0.39	−0.42	0.46

基于遥感与机器学习的中国城市夏季热岛气候带分异特征（2003-2022年）及其驱动因子分析

Spatiotemporal Patterns and Drivers of Summer Urban Heat Islands in Chinese Cities (2003-2022): A Multi-Climate Zone Analysis Using Remote Sensing and Machine Learning

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 11

参考文献 30

相关文章 15

编辑推荐

Metrics

本文评价

气候类型	城市	数量
暖温带	成都，重庆，贵阳，南宁，广州，长沙，南昌，福州，杭州，上海，武汉，合肥，南京，郑州，泉州，漳州，佛山，东莞，宁波，温州，扬州，徐州，蚌埠，平顶山，襄阳，赣州，衡阳，柳州，西安，运城，开封，金华	32
冷温带	沈阳，长春，济南，大庆，鞍山，临沂，齐齐哈尔，唐山，吉林，阜新，锦州，牡丹江，鹤岗，抚顺，秦皇岛，哈尔滨，天津，北京，通辽，赤峰，保定，衡水，石家庄，临汾，邢台，邯郸	26
干带	兰州，银川，太原，大同，包头，乌鲁木齐，呼和浩特	7

作者	城市数量	时间范围	时间变化	空间变化	主要驱动因子
王恭斌等（本研究）	65	2003-2022	夏季白天SUHII以下降趋势为主，夜间以增加趋势为主	夏季白天SUHII在暖温带和冷温带较高，在干旱带城市较低；夜间SUHII相反	昼夜SUHII由△WSA主要影响，不同气候带的其他主要影响因子不同
Zhou et al.，2014	32	2003-2011	-	中国东南部白天SUHII较高，北部夜间SUHII较高	夏季白天SUHII受夜间灯光主导，夜晚受△WSA主导
曹畅等，2017	39	2003-2013	-	白天SUHII在半湿润区城市较高，夜间SUHII在半干旱/干旱区城市较高	夏季白天SUHII由农田像元比例主导，夜晚由△WSA差异主导
刘宇翔等，2022	332	2003-2019	大部分城市昼夜SUHII呈上升趋势；夏季白天上升趋势城市数量高于夜晚	白天SUHII在热带、亚热带及温带的东部城市较高，夜间SUHII在中温带较高	-
Geng et al.，2023	253	2001-2020	大部分城市昼夜SUHII呈上升趋势；白天显著上升趋势分布在南亚热带、中亚热带和北亚热带且占比逐渐增加，夜晚则分布在南温带	白天SUHII在南亚热带城市较高，夜间SUHII在南温带较高	不同气候带昼夜SUHII皆由城市面积主导，不同气候带及季节的其他主要驱动因子不同

[1]	吴锡言, 李维军, 卡那特, 彭玉杰. 基于随机森林和OCO-2遥感数据分析2023年中国连续时空x_CO2变化特征[J]. 生态环境学报, 2025, 34(9): 1341-1350.
[2]	张晟博, 吴作航, 党皓飞, 廖廓, 陆灯盛, 李登秋. 近20年武夷山地区不同遥感指标对极端气候事件的响应及差异[J]. 生态环境学报, 2025, 34(8): 1240-1254.
[3]	祁珣, 冯鑫鑫, 陈颖军, 冯艳丽, 陈田, 李军, 张干. 轻型汽油卡车尾气颗粒物中氨和有机胺的排放特征及影响因素[J]. 生态环境学报, 2025, 34(7): 997-1006.
[4]	张丹丹, 毋振海, 吴渴, 毕方, 李云凤, 安聪, 韩翼昕, 刘正阳, 朱玲, 王学中. 基于不同类型臭氧污染日的臭氧污染特征及影响因素分析：以亳州市为例[J]. 生态环境学报, 2025, 34(5): 720-730.
[5]	李建付, 黄志霖, 和成忠, 姜昕, 宋琳, 刘佳鑫, 陈利顶. 滇东喀斯特断陷盆地土壤有机碳空间分布特征及其关键影响因子[J]. 生态环境学报, 2024, 33(9): 1339-1352.
[6]	戴晓爱, 马佳欣, 唐艺菱, 李为乐. 甘肃省植被时空动态变化及其归因分析[J]. 生态环境学报, 2024, 33(8): 1163-1173.
[7]	李海燕, 杨涛, 廖依琳, 屈亚婕. 渭河流域（陕西段）河流生境质量分布格局及驱动力分析[J]. 生态环境学报, 2024, 33(7): 1153-1162.
[8]	杨非凡, 何浩. 基于“EVI-ESV”伊犁河谷生态环境评估及生态分区构建[J]. 生态环境学报, 2024, 33(4): 655-664.
[9]	李晴, 张梦悦, 于明乔, 李小璇, 常明, 陈立斌, 丁森. 东莞城市河流大型底栖动物群落结构及影响因子[J]. 生态环境学报, 2024, 33(1): 101-110.
[10]	周玉祥, 赵玉, 聂仁东, 丁丁, 郭立华, 周佳峥. 下辽河平原土地沙漠化程度及预测研究[J]. 生态环境学报, 2023, 32(6): 1133-1139.
[11]	董洁芳, 邓椿, 张仲伍. 渭河流域PM_2.5时空演化及人口暴露风险[J]. 生态环境学报, 2023, 32(6): 1078-1088.
[12]	李建辉, 党争, 陈琳. 黄河几字弯都市圈PM_2.5时空特征及影响因素分析[J]. 生态环境学报, 2023, 32(4): 697-705.
[13]	王成武, 罗俊杰, 唐鸿湖. 基于InVEST模型的太行山沿线地区生态系统碳储量时空分异驱动力分析[J]. 生态环境学报, 2023, 32(2): 215-225.
[14]	肖成志, 计扬, 李建忠, 张志, 巴仁基, 曹亚廷. 岷江上游生态脆弱性时空分异及驱动因子交互效应分析——以杂谷脑河流域为例[J]. 生态环境学报, 2023, 32(10): 1760-1770.
[15]	付蓉, 武新梅, 陈斌. 城市地表温度空间分异及驱动因子差异性分析——以合肥市为例[J]. 生态环境学报, 2023, 32(1): 110-122.

时间	显著上升	非显著上升	非显著下降	显著下降
白天	15	29	32	24
夜晚	48	34	11	7

时间	显著上升	非显著上升	非显著下降	显著下降
白天	15	29	32	24
夜晚	48	34	11	7