中文摘要：

地表温度是地球生态环境的重要指标，劈窗算法是目前利用遥感反演地表温度最为成熟的算法。为求解两个通道参数组成的热辐射方程组，需将Planck辐射方程线性展开以使各通道辐射值与亮温Ti直接相关，可通过引入中间参数Li与亮温Ti以不同的数学关系模型进行拟合。以MODIS数据源为例，对4种不同拟合关系进行分析比较，发现线性和幂函数模型具有较高的精度和稳定性，估计Li的误差范围为0．019～0．197。进一步分析这两个模型在地表温度反演中的误差及对参数的敏感性，结果显示幂函数模型在各种情况下都优于传统的线性模型劈窗算法。

英文摘要：

Land surface temperature （LST） is a crucial parameter for earth ecology environment, precise measurement of LST is important to the studies of global change, agriculture, hydrology and meteorology. With the development of remote sensing technology, regional analysis of surface temperature becomes possible. Split window algorithm is considered as the best method for retrieving the LST by remote sensing at present. In order to solve the thermal radiation equation systems combined of two channels＇ parameters, it requires to linearize the Planek radiance equation in order to find the direct correlation of radiance and bright temperature Ti while in this course will bring an interim parameter expressed as Li, thus there may cause some errors even though very small in the processing. Using MODIS as data source, 4 types of fitting relations between interim parameter Li- and bright temperature Ti were compared and analyzed, the results showed both linearity and power models have the best precision and stability in estimating actual Li as well as EST. The preliminary test has been taken on the sensitivity of linearity and power models to atmosphere transmittance and emissivity, and it indicates that the power function model is always better than the traditional linearity split window algorithm.