考虑时空异质性的遥感降水数据降尺度方法研究

The availability of accurate precipitation data with high spatial resolution is deemed necessary for many types of hydrological, meteorological, and environmental applications. A large number of studies demonstrated that traditional point measurements based on meteorological stations cannot reflect the spatial variation of precipitation effectively. The development of remote sensing technology has created an unprecedented opportunity to estimate the spatial distribution of precipitation. However, when applied to local regions, the spatial resolution of these products is too coarse. The issue of downscaling has been listed as one of key areas of research in the application of satellite precipitation data in engineering and making decision. Generally, the existing downscaling methods are still relatively weak. One of the important reasons is that they are unable to consider the relationships which are used to construct the downscaling model to be heterogeneous in both time and space. However, in the real world, these relationships are disturbed under the influence of some factors, such as soil type, hydrological conditions, atmospheric conditions, and human activities. In this study, a novel statistical downscaling method of TRMM 3B43 precipitation considering both time and space heterogeneity is proposed. The basis preprocessing route is as follows: A calibration procedure is performed on annual TRMM precipitation at first; second is performing the spatial downscaling on the calibrated annual TRMM precipitation to attain 1-km annual TRMM precipitation, after which a time downscaling procedure is applied; finally, monthly precipitation at 1-km spatial resolution is produced. The major research contents include: exploring the spatially non-stationary relationships between TRMM precipitation and observation precipitation, and constructing a spatial calibration model of TRMM 3B43 precipitation data; probing into the spatially non-stationary relationships between the calibrated annual TRMM precipitation and the local surface variables, and building a spatial downscaling model of annual TRMM precipitation; detecting the non-stationary time distribution patterns of TRMM annual precipitation by merging time distribution information of monthly precipitation from both TRMM and meteorological stations, and founding a temporal downscaling model of annual TRMM precipitation. This study aims to improve the spatial resolution of remote sensing precipitation data, obtain more accurate downscaling results at the same time, and have a high theoretical and application value.

较粗的分辨率严重制约遥感降水数据的应用，降尺度是解决该问题的一个有效途径。目前针对遥感降水降尺度的研究总体比较薄弱，一个重要原因是它们无法顾及用于构建降尺度模型的时空关系是非平稳的。本研究以TRMM 3B43降水数据为研究对象，提出考虑时空异质性的TRMM 3B43降水降尺度方法。在对TRMM年降水量准确定标基础上，先进行空间降尺度，再进行时间降尺度，最终获得1 km的月降水量。具体包括：研究TRMM降水量和站点观测降水量间的空间非平稳关系，构建TRMM数据定标模型；研究TRMM年降水量和局部地表变量间的空间非平稳关系，构建TRMM年降水量空间降尺度模型；融合TRMM月降水量和站点观测月降水量两种时间分布信息以准确衡量TRMM年降水量非平稳的时间分布模式，构建TRMM年降水量时间降尺度模型。本研究旨在提高遥感降水数据的空间分辨率，并获得精确的降尺度结果，具有较高的理论研究和应用价值。

本项目的主要研究遥感降水数据的时空降尺度，通过对其分析评价获得它的空间分布特性，在此基础上对较粗分辨率的数据进行时空降尺度。具体如下：首先利用地面观测数据对遥感降水数据的质量进行评价，获得其精度在空间上的分布特性，这为接下来的降尺度奠定基础。其次，针对遥感降水精度在空间上的非均匀性，构建了一个考虑空间异质性的遥感降水空间标定模型，相比当面方法，能够大幅度提高遥感降水数据的空间标定精度。最后，提出一种考虑时空异质性的遥感降水时空降尺度方法，融合多源数据来获得高空间以及高时间分辨率的遥感降水数据。共发表SCI论文2篇，EI论文1篇，国家发明专利1项；此外，还有部分成果正在整理中，即将投稿。