基于Ts-NDVI特征空间的绿洲土壤水分监测算法改进

土壤水分胁迫是干旱区绿洲生态环境和可持续发展面临的主要问题，开展区域尺度下大面积、高精度的土壤水分监测，有利于该地区旱情预报、作物估产、气象水文等领域研究。以T_s-NDVI特征空间为理论基础，以新疆渭干河-库车河三角洲绿洲为研究靶区，选择典型干湿季节下Landsat 8遥感影像，在传统温度-植被干旱指数（TVDI）算法基础上，考虑大尺度研究区下垫面异质性（植被覆被、地形起伏）对辐射能量平衡的影响，分别采用植被水分指数（VWIs）、加入大气温度（T_a）和DEM校正后的地表温度（T_s）与NDVI相结合，构建了植被干旱指数（VDI）和改进型温度-植被干旱指数（iTVDI），并结合同期实测土壤水分数据对3种算法进行比较。结果表明：3种算法在一定程度上均能比较客观反映旱情特征，与表层土壤含水量呈现不同程度的负相关，其中，iTVDI相关性最好，TVDI次之，VDI相关性最低；相较植被生长初期而言，3种算法均在植被生长成熟期具有更好的水分监测能力。

Improvement and Comparison of Soil moisture Monitoring Algorithm in Oasis Based on Ts-NDVI Feature Space

This study aims at developing appropriate methods for soil water stress detection in arid regions in Ugan-Kuqa River Delta Oasis by the images of Landsat 8. To do this, we use NDVI combined with Vegetation Water Index (VWIs) and surface temperature (T_s) to construct Vegetation Dryness Index (VDI) and Temperature Vegetation Drought Index (TVDI) respectively. At the same time, a modified approach towards the TVDI incorporating air temperature (T_a) and a DEM to develop the improved Temperature Vegetation Drought Index (iTVDI), which taking into account the impact of topography and cover types. The three algorithms were applied to retrieve the spatial and temporal distribution of water stress. Then, the same period field surface soil moisture data were used for verification and evaluation. The results show that the three algorithms to some extent, all can objectively reflect the dryness characteristics and all have negative correlation of soil moisture. While iTVDI has best correlation, TVDI followed, VDI has minimum correlation. In addition, all R² values in April were lower than values in August, it is concluded that compared with growing season, the three algorithms were more suitable for the grown season water stress/drought detection.