多源地面与遥感观测联合的径流数据同化方法研究

Multi-souce data assimilation is one of the leading directions of hydrological data assimilation research at present. It is rooted in the theory frame of data assimilation, combines hydrological model with multi-sourve ground and remote sensing observation data, and obtains the hydrological information which accords with the law of surface water balance and dynamic mechanism, and contains a variety of effective observation informations. In this study, the Babaohe River Basin in the upper reaches of Heihe River Basin is studied, a data assimilation system is built with Weather Research and Forecasting Model (WRF) model and grid Soil and Water Assessment Tool (SWAT) model and the precipitatoin and soil moisture data of distributed ground observation and remote sensing observation are respectively and jointly assimilated. The method of data assimilation for distributed soil moisture data assimilation and jointly assimilation of precipitation and soil moisture observations is developed. The research ideas and methods will further enrich and perfect the theory system of streamflow data assimilation, and the combined data assimilation method of soil moisture and precipitation observation provides a new idea for multi-source hydrological data assimilation, and also provides a basic framework for multi-source multiscale data assimilation research.

多源数据同化是目前水文数据同化研究的一个前沿方向，它植根于数据同化的理论框架内，将水文模型与多源地面和遥感观测资料进行有效地融合，得到既符合地面水量平衡规律和动力学机制、又包含多种有效观测信息的水文资料。本研究准备以黑河上游八宝河流域为研究区，耦合WRF模型与网格化SWAT模型，搭建同化系统，并在此基础上分别和联合同化分布式地面观测和遥感观测的降水与土壤水分数据，发展分布式土壤水分数据同化方法和降水与土壤水分观测数据的联合同化方法。研究思路和方法将进一步充实和完善径流数据同化的理论体系，土壤水分与降水观测的联合数据同化方法为多源水文数据同化提供了新的思路，也为多源多尺度的数据同化研究提供了框架基础。

径流的模拟及预报工作对于区域的水资源管理与合理利用、灾害预警预报、生态环境保护以及经济发展等均具有着重要的科学意义，但单纯的观测与模拟难以提供高精度的径流预报数据，所以迫切需要发展能够结合多源观测信息与模拟信息的径流数据同化技术。本项目以我国西北内陆河流域黑河上游八宝河流域为研究区，利用流域数据同化系统同化遥感与站点观测的降水和土壤水分数据， 分析径流模拟精度的改进结果。 主要研究内容包括：将SWAT模型进行网格化处理，使网格化的水文响应单元与WRF网格单元对应，变量间的传递工作通过文件传输完成，实现WRF模型与SWAT模型的离线耦合模式。然后我们将并行同化框架PDAF与WRF-SWAT整合，搭建了WRF-SWATDAS框架。利用WRF-SWATDAS同化系统同化分布式土壤水分数据来提高径流模拟精度，结果表明分布式土壤水分数据同化可以提高流域整体土壤水分和径流估计精度；利用SM2RAIN算法将土壤水数据反演为降水数据，再联合利用反演的与站点观测的降水数据来驱动水文系统进行水文模拟，结果表明土壤水分可以有效反演降水数据，并进行径流预报。在WRF-SWATDAS框架下，学者们可以综合利用WRF模型、SWAT模型、多种数据同化方法，多源观测数据（站点土壤水分数据、遥感观测的降水数据），改进径流模拟结果。