考虑模型背景场与多源观测值自相关结构差异的根层土壤水分同化研究

项目名称： 考虑模型背景场与多源观测值自相关结构差异的根层土壤水分同化研究

项目编号： No.41501450

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 天文学、地球科学

项目作者： 邱建秀

作者单位： 中山大学

项目金额： 20万元

中文摘要： 融合多源、多分辨观测资料的数据同化方法是准确预报根层土壤水分的有效手段，为当前研究热点。数据同化过程中模型背景场与多源观测值之间的时域自相关结构差异将降低同化系统性能，显著影响根层土壤水分的预报精度。而已有的研究对于该自相关差异及其对同化结果的影响鲜有关注，是同化研究中的薄弱部分。本项目以华北平原崇岭流域为研究对象，拟设计一系列理想合成实验与真实同化实验，系统地分析在不同表-根层耦合强度下，陆面同化系统对不同来源的自相关差异的敏感性，定量化评估该自相关差异的影响。在此基础上构建有效的诊断指标体系，鉴别并校正同化系统中的自相关差异，从而避免这一系统误差造成同化性能的降低，进而提高研究区域的根层土壤水分预报精度。本研究的结果将有效扩充陆面同化研究理论体系，并为进行区域农业干旱监测预警提供科学保障和技术支持。

中文关键词： 土壤含水量；遥感水文耦合模型；主动微波遥感；被动微波遥感；雷达遥感

英文摘要： Data assimilation, which incorporates multi-source and multi-resolution observations, is an effective technique for root-zone soil moisture (RZSM) forecast, and thus the hot research topic in hydrological community. During assimilation, the mismatch in temporal autocorrelation structure between observations and comparable model-based predictions will undermine the efficiency of such assimilation and significantly reduce the accuracy of RZSM prediction. However, there is rare investigation focusing on this mismatch and its impact on data assimilation system. This research aims to quantify the impact of model/observation auto-correlation mismatch on RZSM assimilation through a series of synthetic and real data assimilation experiments. As the autocorrelation mismatch can be attributed to both of the following effects: 1) vertical uncertainty in the microwave emission model and 2) error in the timescale of soil moisture dynamics produced by the land surface model, different experiments are designed to capture the subsequent impact of different sources of model/observation autocorrelation misfit. Also, given that autocorrelation structure varies with the manner of how soil moisture in each model layer are weighted, it stands to reason that the sensitivity of data assimilation results to autocorrelation misfit will vary as a function of vertical soil moisture coupling strength. Therefore, this research also carefully examines the impact of autocorrelation mismatch with various vertical coupling strengths. The second objective of this research is to propose a robust diagnostic index system which examines the auto-correlation characteristic of model predictions and observations, and thus discern the presence of auto-correlation differences. Application of this diagnostic index should allow for the identification of worst-case scenarios in which auto-correlation differences are largest and significantly degrade assimilation performance. This research will supplement the land data assimilation theories, and research results will be valuable in agricultural drought warning practice at regional scale.

英文关键词： soil moisture content;coupling of remote sensing and hydrological model;active microwave remote sensing;passive microwave remote sensing;radar remote sensing