GNSS极化掩星信号反演雨团垂直结构方法

Method for retrieving vertical structure of rain clusters by GNSS polarimetric occultation signals

本文紧跟国际上新兴起的GNSS提取降雨信息这一研究思路,重点针对利用GNSS极化掩星信号提取雨团垂直结构问题进行了建模、解算和仿真研究.以GNSS-LEO掩星事件为背景,基于GNSS掩星信号经过雨团产生极化相移的物理特性,借鉴计算机层析技术的原理,构建了层析雨团垂直结构的数学物理模型,并提出了采用两种反演算法(TSVD法、Tikhonov法)及两种正则化参数选取方法(L曲线准则、GCV准则)的反演方案;通过数值模拟,分别研究了该方案在超定和欠定情况下的可行性,并对比分析了不同反演算法的效果.结果表明:超定情况时,上述算法均能较准确地重构出雨团垂直结构,和原始分布的相关系数超过0.96,且不同算法采用不同正则化参数选取方法时结果相同;欠定条件下,甚至在未知数个数远大于观测量个数时,该方案仍可重构出雨团信息,不同仿真试验最优结果的相关系数位于0.75~0.98之间,且Tikhonov法明显优于TSVD法.该研究对利用GNSS信号获取区域雨团信息进行了有益探索,为后续低轨卫星载荷方案设计和论证提供了技术支撑.

Following the new international research idea of extracting rainfall information by GNSS,modeling,calculation and simulation of vertical structure of rain clusters are studied by using GNSS polarimetric occultation signals.Based on the physical characteristics of polarimetric phase shift induced by GNSS occultation signals passing through precipitation during a GNSSLEO occultation event,a physicalmathematical model for the vertical structure retrieve of rain clusters is constructed using Computer Tomography for reference;a scheme with two algorithms(TSVD method and Tikhonov method)using two regularization parameter selection methods(L-curve criterion and GCV criterion)is proposed to solve this model.Finally,through numerical simulation,the feasibility of this scheme in the case of overdetermined and underdetermined conditions is studied,and the effects of different inversion algorithms are compared and analyzed.The results show that in the case of over determination,the proposed inversion scheme can reconstruct the vertical structure of rain clusters quite accurately with the correlation coefficients more than 0.96,and the four results of the above two regularization algorithms using two regularization parameter selection methods are the same;under the condition of underdetermination,even when the number of unknowns is much greater than the number of observed quantities,this scheme still has the feasibility to get main information about rain cluster structure with the correlation coefficients of the optimal results at different simulations between 0.75 and 0.98,and the Tikhonov method is better than the TSVD method.This research makes a meaningful exploration for obtaining regional rain cluster information from GNSS signals,and provides technical support for the design and demonstration of the following LEO satellite payload scheme.