减化比值植被指数在城市热环境研究中的应用

以南京市ETM＋影像为数据源,经嵌套于ERDAS IMAGINE9.1中的ATCOR2大气校正后,提取了南京城区地表温度（land surface temperature,LST）、归一化植被指数（normalized difference vegetation index,NDVI）和减化比值植被指数（reduced simple ratio,RSR）,基于实测值和同时相AS-TER数据反演结果的双重验证,拟合了两种植被指数与地表温度的定量关系,并进行对比分析。结果表明,除水体外,植被覆盖度高的区域地表温度明显低于植被稀少或无植被区;地表温度与NDVI呈显著线性负相关,与RSR呈显著幂函数负相关,后者的相关性高于前者;当RSR小于3.2时,地表温度随植被覆盖增加而锐减,当RSR大于3.2时,植被覆盖继续增加,地表温度却趋于恒定,呈现植被降温效应＂饱和＂现象;对比比值植被指数（ratio vegetation index,RVI）应用结果发现,近红外和红光比值的线性拉伸是导致NDVI和RSR表征城市热环境效应存在差异的主因,还与短波红外对水分敏感和能真实反映植被冠层结构有关。中国6大城市热环境案例研究均表明,RSR能直接解释对于不同下垫面,增加相同的植被覆盖度其降温效果存在差异的现象,而不进行土地利用分类,NDVI则无法揭示此现象。

Application of reduced simple ratio in urban thermal environment research

Based on the ETM＋image,the land surface temperature（LST）,reduced simple ratio（RSR） and normalized difference vegetation index（NDVI） of Nanjing are retrieved after effective atmospheric correction by ATCOR2 nested in ERDAS IMAGINE 9.1.In comparison of ETM＋LST with ASTER LST and observed values,it shows that good consistency exists both at single station and in spatial distribution.Regression and contrastive analysis are conducted for explaining relationships between LST and the two vegetation indexes.Results show that the LST associating with thick vegetation is much lower than that with sparse vegetation or bare lands,when water bodies are without consideration.Fitting models prove a significant negative linear correlation between LST and NDVI,while LST and RSR are more obviously correlated with power function.When RSR is below 3.2,LST decreases rapidly with the vegetation coverage increasing.However,it tends to be unchangeable when RSR exceeds 3.2.Cooling effect of vegetation is at ＂saturation＂,and only RSR can reveal the interesting phenomenon.Compared to the ratio vegetation index（RVI）,the linear ratio of near infrared and red is suggested as the main reason for different behavior of the two vegetation indexes applied in urban thermal environment.Other factors may include better sensitivity to moisture and canopy closure information detected by near infrared.Urban thermal environment research in six cities of China demonstrates that cooling effect of vegetation is limited by underlying surface patterns.Compared to NDVI,RSR is not only better for indicating the relationship between LST and vegetation coverage,but more effective for indicating thermal environment on city scale.