土壤水是维持地球生命的重要自然资源，准确获取土壤含水量变化信息对于维持地球生态系统平衡具有十分重要的意义。目前在1km、逐日的高时空分辨率尺度上，遥感技术还难以获得对任意大范围区域实现稳定全覆盖的土壤水分产品。本研究目标即以中国为研究区，以融合光学遥感数据源的被动微波土壤水分降尺度技术为基础，研发中国首套1km分辨率逐日时空全覆盖遥感地表土壤水分产品。针对降尺度所需的光学（热红外）LST在多云时易产生缺失像元的问题，研究将改进现有的LST时空融合插补方法并得到空间全覆盖的热红外LST。之后，将以地面大田蒸发试验为基础，构建用于估算传统土壤水分降尺度方法的适用边界阈值的数学模型；对于土壤水分含量高于边界阈值的极湿地表，将通过引入机器学习方法改进传统降尺度模型，从而建立对全湿度范围适用的高时空分辨率土壤水分估算方法体系。估算结果将与不同站点观测数据进行对比验证，最终产品将在公共数据平台发布。

Soil moisture is an important natural resource for the existences of earth lives. Accurate estimation of land surface soil moisture is of great significance on maintaining the global ecological system balance. Currently, however, it is still difficult to obtain stable and full-covered soil moisture products with high spatio-temporal resolutions like 1 km on a daily basis when applying satellite remote sensing techniques in an arbitrary large-scale region. On this premise, this study will take the entire China as study region, and develop a novel methodology framework for generating the first set of full-cover surface soil moisture product all over China at 1 km on a daily basis, based on the passive microwave soil moisture downscaling technique through fusing optical data sources. Since one of the primary optical data (including "thermal infrared band" here) sources, i.e. land surface temperature (LST), can suffer from pixel loss easily over cloudy areas, we will firstly build a model to improve traditional LST interpolation methodology, through which a full-covered daily LST dataset at 1 km will be generated. In the next step, we will conduct a series of field-scale experiments for simulating the soil evaporation process. Through the experiments, we aim to develop a mathematical model which can well estimate the boundary soil moisture threshold suitable to traditional soil moisture downscaling methods. And we will introduce several machine-learning-based models to improve the traditional downscaling methods when it is under an extremely wet surface with soil moisture values higher than the aforementioned boundary threshold. In this way, we propose to establish an integrated methodology framework that is suitable to a much more complete variation range of soil moisture values. Model estimates will be evaluated against in-situ soil moisture observations and the final soil moisture product will be published in an authoritative public data plateform.