中文摘要：

结合Landsat 8数据特征提高遥感反演土壤水分的精度,对旱灾的监测具有重要意义。主要是利用Landsat 8数据,基于植被指数（NDVI）-地表温度（T_s）特征空间理论,通过对植被指数和“干边”的修订,分别计算温度植被干旱指数（TVDI）、干边修订的改进型温度植被干旱指数（MDTVDI）;并结合当地气象数据、野外实测土壤水分数据,对计算结果进行了对比。结果表明：在使用Landsat 8OLI和TIRS数据时,TVDI,MDTVDI与地表土壤水分线性相关显著（α=0.05）,两者都能表征地表的干旱分布情况;同时,MDTVDI与实测土壤水分的相关性和反演的土壤水分精度略高于TVDI,因此MDTVDI能够更好地反映区域土壤水分状况,更适于区域干旱监测。

英文摘要：

It is significant to improve the retrieval accuracy of soil moisture for the drought monitoring based on the features of Landsat 8 data. In this paper, based on the feature space theory of vegetation index （NDVI） - surface temperature （Ts）, temperature vegetation drought index （TVDI） and modified dry- edge -revised temperature vegetation drought index （MDTVDI） were respectively calculated using Landsat 8 data. To do this, the NDVI and ＂dry-edge＂ was revised; and then the retrieval results were compared with the local meteorological data and field measured soil moisture data. Results show that, Landsat 80LI and TIRS data performed well in retrieving soil moisture. There are significant linear correlations（α = 0. 05） between TVDI, MDTVDI and surface soil moisture. Both TVDI and MDTVDI can characterize the distribution of surface drought well. In addition, the correlation be- tween MDTVDI and the measured soil moisture was slightly higher than that of TVDI, besides, the retrieval accura- cy of soil moisture using MDTVDI was also slightly higher than that using TVDI. Compared with TVDI, MDTVDI could reflect the regional soil moisture states better and thus was more suitable for regional drought monitoring.