基于ETM＋遥感影像反演不同土地利用类型地表温度的研究

采用Landsat7 ETM＋为基本数据源，运用3种算法（大气辐射传输方程RTE、Qin等单窗算法和Jimenez—Munoz＆Sobrino普适性单通道算法）定量反演了黄河三角洲部分地区的地表温度（land surface temperature，LST），并进行了不同算法反演结果的差值比较：以RTE反演结果为标准，在大气水汽含量较低时，Qin等单窗算法和JM＆S普适性单通道算法精度较高，与RTE的反演结果相差均在1K以内；在大气水汽含量较高时，Qin等单窗算法在采用地面气象资料估算水汽含量条件下仍保持较高的精度，比RTE反演结果平均偏小0．95K；而根据估算的大气水汽含量进行反演的JM＆S普适性单通道算法的反演偏差比Qin等单窗算法要大，达到1．94K，但JM＆S普适性单通道算法根据实测的大气水汽含量得到的反演结果要略优于Qin等单窗算法，比RTE结果偏大0．67K。同时计算了经过6s模式校正后归一化植被指数（INDV），然后利用GIS中的空间分析功能，分析LST、INDV在不同土地利用类型之间的差异以及二者之间的定量关系。发现研究区内，就所有土地利用类型而言，平均LST和INDV之间存在显著的负相关关系；对于各种土地利用类型，这种相关关系也存在．但相关程度不同。

Land Surface Temperature Retrieval of Different Land Use Types from ETM+ Images

The land surface temperature （LST） of the study area of the Yellow River Delta are retrieved from Landsat7 ETM ＋ images using three different methods （ thermal radiance transfer equation, the mono-window algorithm by Qin et al in 2001, and the generalized single-channel method by JimenezMunoz ＆ Sobrino in 2003 ） , and the mean accuracy of retrieval results also disscussed. The retrieval resuits show that the mono-window algorithm and the single-channel method both have a higher accuracy within 1 K deviation when the integrated water vapor in the atmosphere is low; While the water vapor of air is high and the LST is retrieved with the water vapor estimated by the ground-base weather data, the error is about 0.95 K for the mono-window algorithm in contrast to 1.94 K for the single-channel method. Meanwhile,the Normalized Difference Vegetation Index （INDV） is corrected by 6S mode, and then, differences in LST and INDV for different land use types and their quantitative relations are analyzed. The result of regressive analysis shows a significant negative correlation relationship between LST and INDV for all land use types, and the negative correlation holds for each land type, but with different correlation coefficients.