基于矩阵填充理论的土壤湿度地面观测数据稀疏性问题研究

目前，研究高精度、高空间分辨率的土壤湿度数据已成为干旱监测、农业应用、气候模拟的迫切需求。本项目探索研究基于矩阵填充理论用于提高土壤湿度地面观测数据插值方法精度和效率的相关机理，通过剖析矩阵填充理论中空间数据的低秩性、非相干性、完整性，研究SVT矩阵重构算法，从理论上建立基于矩阵填充理论的空间数据稀疏性插值重构模型；深入研究土壤湿度地面观测数据的空间分布特性，分析数据之间的信息相关性，针对土壤湿度数据空间 变异大的特点，研究未知重构区域的空间变异程度，构建可准确表征时空分辨率高土壤湿度数据的矩阵填充模型。本研究为土壤湿度地面观测数据稀疏性问题提供良好的理论指导，对促进 其在陆面数据融合和同化中的应用具有重要的实际应用价值。

At present, soil moisture data of high spatial resolution and high-precision has become the urgent need in terms of drought monitoring, agricultural application and climate simulation. The project explores the correlation theory based on the Matrix Completion for improving the interpolation precision of automatic soil moisture station data. By analyzing spatial data of the low rank, incoherence and integrity of the matrix theory，studying SVT matrix reconstruction algorithm, we can establish sparse interpolation of the spatial data reconstruction model based on matrix filling theory; in-depth study of automatic soil moisture station data spatial distribution feature, thereby analyze the correlation between the data information. As for characteristics of soil moisture data of large spatial variability, we study the degree of spatial variability of the unknown reconstructed region and build a matrix completion model that can accurately characterize soil moisture data of high spatial resolution. The project provides a good theoretical guidance for studying automatic soil moisture station data sparsity problem and it has important practical value to promote its use in land data fusion and data assimilation.