基于多源数据的对流层水汽层析建模及其稳定求解算法研究

Water vapor inversion using multi-source data is a hot topic in the research of water vapor tomography. The existing water vapor inversion has an ill-posed problem caused by the low utilization rate of inherent multi-GNSS data and the weak strength of satellite-station geometry structure, and there is a lack of relevant algorithms for determining the optimal weighting of multi-source data and stable solving of the tomographic model. In this project, the mixed truncation factor is introduced to study the functional relationship between the factor and the water vapor content in the tomographic area, and the three-dimensional water vapor tomography algorithm is established to improve the utilization rate of multi-GNSS observations and the stability of the tomographic normal equation. Based on the idea of co-integration test and the improved variance component estimation method, the weighting of different types of data in the tomography model is evaluated, and the optimal weighting determination method of multi-source data suitable for the water vapor tomography model is developed to provide accurate weighting information for the stable solution of the tomography model. Based on the regularization theory, a new approach to determine the optimal regularization parameters by using the principle of minimum deviation and the idea of iterative optimization is explored，and the reconstruction of tropospheric water vapor tomography modeling based on multi-source data and its stable solution are realized. It lays a theoretical and applied foundation for the early warning of extreme weather and the improvement of the initial field of numerical prediction model, and promotes the development of water vapor tomography technology in the field of meteorology.

多源数据水汽反演是目前水汽层析研究中的热点问题。现有水汽反演中存在因固有multi-GNSS数据利用率低、星站几何结构强度弱等导致的不适定问题，层析模型解算时缺乏多源数据最优权值确定和模型稳定求解的相关算法。本项目引入混合截断因子，研究该因子与层析区域内水汽含量之间的函数关系，建立提高multi-GNSS观测数据利用率的三维水汽反演算法，改善层析法方程的稳定性；联合协整检验思想和改进的方差分量估计方法，评估不同类型数据在层析模型中的权重，发展适用于水汽层析模型的多源数据最优权值确定方法，为层析模型的稳定求解提供精确的权比信息；基于正则化理论，探索利用最小偏差原则和迭代优化思想确定最佳正则化参数的新途径，提出针对多源数据层析模型的正则化稳定求解方法，实现基于多源数据的对流层水汽层析建模及其稳定求解。为极端天气预警和改善数值预报模式初始场奠定理论和应用基础，推动水汽层析技术在气象学领域的发展。

高精度、高时空分辨率水汽信息对于天气和气候变化等研究具有重要意义，项目根据多源水汽层析反演中存在GNSS数据利用率低、多源水汽信息权比不确定等难题开展研究。针对GNSS观测数据利用率低的问题，提出一种顾及层析区域外测站数据的水汽层析方法。该方法通过引入比例系数并联合水汽密度初值建立比例系数模型，然后进一步确定层析区域外测站上的观测数据在层析区域内的水汽含量，构建相应的水汽层析观测方程，用于改善水汽层析法方程的稳定性。针对多源水汽融合反演中不同数据源权比不确定的难题，提出一种混合水汽融合方法，联合球谐函数与多项式拟合对水汽偏差进行模型表达，并进一步利用Helmert方差分量估计自适应确定多源水汽信息权比，量化多源数据对水汽融合的贡献。针对传统水汽层析模型解算时小奇异值的存在会导致层析结果存在较大波动的现状，提出一种改进岭估计的水汽层析模型解算方法，在现有正则化方法的基础上，附加最小偏差约束对现有岭估计方法进行改进，只对层析结果有严重影响的小奇异值进行处理，最终实现水汽层析模型的稳定求解。此外，项目研究进一步拓展了多源水汽信息在短临降雨预警、干旱监测指数改进与构建、AOD建模与监测方面的创新应用，对于GNSS气象学在天气和气候变化等方面研究具有一定的促进意义。

内容获取失败，请点击重试