Most infrared satellite remote sensors have a higher spatial resolution than microwave satellite sensors. Microwave satellite remote sensing has proven successful for the retrieval of soil moisture (SM) information. In this study, we propose a SM retrieval algorithm based on temperature vegetation dryness index (TVDI), a function of land surface temperature (LST), and the normalized difference vegetative index (NDVI) provided by Moderate Resolution Imaging Spectroradiometer (MODIS) data. We implemented the LST correction with elevation effect. Conversion relationships between TVDI and SM content for a variety of land types were obtained from spatial and temporal collocation of TVDI and Global Land Data Assimilation System (GLDAS) SM content for 2014. From the comparison with the GLDAS SM for 2015, the proposed TVDI-based SM algorithm showed good performance with CC = 0.609, bias = −0.035 m³/m³, and root-mean-square-error (RMSE) = 0.047 m³/m³, while the Soil Moisture Active Passive (SMAP) and Soil Moisture and Ocean Salinity (SMOS) SMs present CC = 0.637 and 0.741, bias = 0.042 and 0.010 m³/m³, and RMSE = 0.152 and 0.103 m³/m³, respectively. For the in situ SM measurements of the Korea Rural Development Administration (RDA), the proposed TVDI-based SM algorithm yielded CC = 0.556, bias = −0.039 m³/m³, and RMSE = 0.051 m³/m³ excluding the winter season. Consequently, the proposed SM algorithm could contribute to complementing the low spatial resolutions of microwave satellite SM products and low temporal resolutions of GLDAS SM products.

中文翻译：

---

基于温度植被干度指数和中分辨率成像光谱仪数据的红外土壤水分反演算法

大多数红外卫星遥感器比微波卫星传感器具有更高的空间分辨率。事实证明，微波卫星遥感技术能够成功检索土壤水分（SM）信息。在这项研究中，我们提出了一种基于温度植被干燥指数（TVDI），地表温度（LST）函数和中分辨率成像光谱仪（MODIS）数据提供的归一化植物营养指数（NDVI）的SM检索算法。我们实施了具有高程效应的LST校正。根据2014年TVDI与全球土地数据同化系统（GLDAS）SM内容的时空搭配，获得了各种土地类型的TVDI和SM内容之间的转换关系。通过与2015年的GLDAS SM进行比较，³ / m ³和均方根误差（RMSE）= 0.047 m ³ / m ³，而土壤水分主动-被动（SMAP）和土壤水分与海洋盐度（SMOS）的SM分别为CC = 0.637和0.741，偏差＝ 0.042和0.010m ³ / m ³，RMSE ＝ 0.152和0.103m ³ / m ³。对于韩国农村发展局（RDA）的现场SM测量，建议的基于TVDI的SM算法得出CC = 0.556，bias = -0.039 m ³ / m ³和RMSE = 0.051 m ³ / m ³不包括冬季。因此，所提出的SM算法可以有助于补充微波卫星SM产品的低空间分辨率和GLDAS SM产品的低时间分辨率。