多维地基GNSS水汽反演与数值同化及其在对地观测技术中的应用

The spatio-temporal variation of water vapor causes various weather phenomena and also influences the accuracy and reliability of Earth observation techniques. Water vapor tomography technique makes the dynamically and multi-dimensionally monitoring of spatial water vapor achievable. However, the water vapor tomography technique still faces some problems: 1) for flat GNSS networks, conventional water vapor tomography algorithms have a strong dependence on a priori information and the accuracy of tomographic results significantly relies on the accuracy of a priori information; 2) tomography products haven’t been effectively used in fields like meteorology, etc. To solve the tomography problem, this project plans to develop an a priori information free tomography algorithm based on adaptive smoothing constrain and Helmert variance component estimation. This algorithm should have an accuracy equivalent to the conventional ones without any a priori information and the accuracy can be further enhanced by a few surface meteorological data. To solve the problem of using tomography products, this project will investigate the assimilation of multi-dimensional GNSS water vapor (delay) products by WRFDA model and the produce of reanalysis and numerical forecast products by assimilating multisource data including GNSS water vapor (delay) products, ground meteorological data and radiosonde data, etc. To enhance the application of the products, the project will investigate the solution of using the produced products to correct the tropospheric delay in Earth observation techniques, like GNSS and InSAR, etc.

水汽的时空变化会产生各种天气现象并会影响对地观测技术的精度和可靠性，GNSS层析技术可实现空间水汽的多维动态监测。但层析技术还存在以下问题：1)对于扁平的GNSS观测网，常规水汽层析算法对先验信息有着严重依赖，层析结果取决于先验信息的精度; 2)层析产品在气象学等领域没有得到有效利用。为解决层析算法的问题，本项目拟基于自适应平滑技术和赫尔默特方差分量估计发展一种不依赖先验信息的水汽层析算法，该算法在无先验信息时能够取得与常规算法相当的精度，在有少量地表气象数据时，精度能够进一步增强。为解决层析产品的应用问题，研究利用WRFDA模型同化多维GNSS水汽(延迟)产品的技术方案，发布基于区域GNSS数据、地表气象数据和无线电探空数据等多源数据同化的再分析和数值天气预报产品。为了提高成果的利用率，研究利用上述产品对GNSS和InSAR等对地观测技术进行对流层延迟改正的技术方案。

水汽在一系列尺度的天气现象和气候变化中起着重要作用，它的时空变化会影响大气的垂直稳定度。对水汽进行精细监测可提升天气预报和气候变化监测水平。然而，当前的水汽观测手段还存在诸多不足，GNSS技术的发展为水汽监测提供了一种全新的手段。本项目构建了高精度高分辨率的中国区的对流层延迟和加权平均温度模型，提升了GNSS监测二维水汽的精度；提出了基于自适应平滑约束和赫尔默特方差分量估计的水汽层析算法，解决了GNSS三维水汽反演中方程不适定的问题；开展了GNSS对流层延迟/水汽产品的数值同化实验，并与GNSS水汽层析技术得到湿折射率进行了比较，结果表明同化GNSS水汽产品可以改善数值天气预报产品的精度，层析技术与数值同化技术在获取水汽上的精度相当；提出了基于球冠谐模型和赫尔默特方差分量估计的多源水汽数据融合算法和基于人工神经网络模型的融合算法，两种方法都取得了较好的效果。本项目的主要成果解决了GNSS气象学中的关键问题，包括二维/三维水汽监测问题、水汽数据的同化问题以及多源水汽数据的融合问题，拓展了GNSS技术的应用空间，对于大地测量学和气象学的发展都具有一定的促进意义。