Abstract: With Landsat 8 two images of Kunming urban area as the data source, normalized difference vegetation index (NDVI), modified normalized water index (MNDWI), impervious surface area (ISA) and thermal field variation index to reflect the change of heat island effect in different seasons were carried out. The results showed that: (1) Using the improved single window-algorithm to calculate the land surface temperature (LST) was effective. (2) The heat island effect existed spatio-temporal differences, and there was stronger effect and more concentrated distribution in May than in February. And the heat island effect level had a close relationship with land use types, building type with the strongest heat island had the maximal proportion in two months, the proportion of covering with water with strongest heat island was minimum. The燚ian燣ake had no heat island effect for its big water area. (3) The correlation between LST and surface parameters in February and May had the same trend, significant positive correlation with the ISA, and significant negative correlation with NDVI and MNDWI. The correlation between LST and NDVI, ISA in May were higher than in February, while the correlation between LST and MNDWI in February was higher than in May. (4) The urban green space and water could cool the surface temperature of the city, protecting the Dian Lake and increasing the vegetation coverage moderately will play a positive role in suppression of the urban heat island effect.

Key words: urban heat island, impervious surface area, NDVI, MNDWI, Kunming

摘要： 以Landsat 8 两期昆明市主城区影像为数据源，通过计算归一化植被指数(NDVI)、改进的归一化水体指数(MNDWI)、不透水面指数(ISA)以及热场变异指数反映不同季节热岛效应的变化情况。结果显示：(1)利用改进的单窗算法反演地表温度(LST)是有效的；(2)热岛效应存在时空差异性，5月的强效应等级比2月更明显，分布也更集中，热岛效应等级与土地利用类型关系密切，强热岛效应分布于建筑用地在2月和5月均较大，而水体所占比例最小，在大的水域滇池则无热岛效应，其他地类在不同季节的强度存在差异；(3)LST与地表参数的相关性在2月和5月表现出相同的趋势，与ISA具有显著的正相关关系，与NDVI和MNDWI具有显著的负相关关系，LST与NDVI、ISA的相关性5月高于2月，而与MNDWI的负相关性2月高于5月；(4)城市的绿地和水体对其地表温度起到降温作用，保护好滇池和适度增加植被覆盖对抑制城市热岛效应具有积极作用。

关键词: 城市热岛, 不透水面率, NDVI, MNDWI, 昆明