基于多源大气观测约束的北斗-GPS三维水汽层析技术研究

Based on the present methods such as Kalman filter and SVD (Singular Vector Decomposition)，three-dimensional water vapor field can be retrieved by the based ground GPS (Global Positional System)observations but have the low accury values on the region edge because of the rank lack in solving the mathematical equations and the low resolution by the sparse observation network.Multi-source observations being used to improve the GPS-Met water vapor retrieved method is more reasonable than the pure mathematical algorithms. To resolve the above problem in this project, the multi-source observations can be used as the additional water vapor observational constraints conditions: the vertical constraints condition is made by the radiosonde obseving data ,the horizontal constraint is made by the GAUSS Function,and the ground observations can be used to add the observation in the low level of the reserch domain. Ninteen satellites working on the orbits and their 30 receiving stations working on the ground, BeiDou (BeiDou Navigation Satellite System) can be used in BeiDou-GPS water vapor observing. The PANDA software made by Wuhan University and the GAMIT software by MIT are used to preprocess the BeiDou and GPS data respectively and build the observation equation of the 3D Water Vapor Tomography. Based the optimum interpolation method and the mathematical method, these multi-observations can be used to improve the retrieval accuracy of atmospheric water vapor, especially the accuracy of values in the grid edge. The retrieve results were tested by comparing with the direct observation by the radiosond, and the accuracy of the retrieval results in the 3D tomography domain of the overall, the low-level, the center and the edge will be analyzed. By the mature variation technique of LAPS(Local Analysis Precipitation System), the retrieve results will be integrated to join the mesoscale numerical model forecasting.the model foecasting results such as the humidity and precipitation with and without GPS-Met 3D vapor will be compared with each, and the rainfall forecasting results will be compare with the fact precipitation.

关于地基GPS三维水汽场反演，国内外主要采用数学理论如Kalman滤波或SVD奇异值分解法解决观测方程组缺秩问题，但网格边缘或站点稀疏区域反演数据常常不真实，若加以大气观测约束条件解决方程组缺秩则比较合理。我国北斗卫星导航系统已具备水汽观测条件急待有技术同步。本项目拟分别采用PANDA软件和GAMIT软件开展北斗-GPS导航系统数据的前处理研究，得到斜路径水汽总量；由斜路径水汽在层析网格中的分布建立三维水汽密度观测方程组，基于多年探空直接观测作为垂直约束，以高斯函数作为水平约束，以地面观测楔入增加层析区域低层观测值等方法建立新观测方程组，解决缺秩问题，最终提供区域三维水汽密度及其与探空观测秒数据的对比检验结果。新算法预期改善边缘和低层的反演精度，北斗-GPS的联合观测技术将改善因观测网稀疏对三维水汽层析的影响，推进北斗导航卫星系的水汽观测技术，为中尺度数值预报模式与天气分析提供精细水汽场。

水汽是形成降水的必要条件之一，高精细的三维水汽观测数据不仅是深入揭示小尺度灾害性天气机理的宝贵资料，对提高降水数值预报也至关重要。因受观测限制，卫星、探空等传统观测方法获得三维水汽场分布在时间和空间分辨率上很难达满足中尺度暴雨研究与预报的要求。北斗卫星导航系统（简称BDS）与GPS大气水汽探测原理相同，能高时空分辨率、全天候连续地开展观测，遥感水汽的前处理技术已经成熟，有必要利用这种新型的遥感探测获得高精细的三维大气水汽信息。.本项目以长江中游鄂东为研究区域，依据区域内GPS站22个，BDS站17个，以及1部微波辐射计、1个探空站和若干地面站，开展关于地基BDS-GPS联合使用与三维水汽场反演技术研究：北斗、GPS导航系统延迟量ZTD基本一致，在6天的3503个样本中，差值小于50 mm的样本比例为97.3%， 2种水汽观测与探空相关系数大于0.95，平均偏差0.6-5mm之间。具备BDS-GPS联合大气水汽探测条件；建立了北斗、GPS的斜路径水汽SWV观测算法；并基于SWV引入3年以上的探空直接观测作为垂直观测约束，以高斯函数作为水平约束，以地面观测增加低层观测值，建立区域BDS-GPS三维大气水汽反演方程组，得到了水平分辨率30km垂直分辨率1km的高分辨率三维水汽密度场，与格点结果与同址探空观测比较，相关系数0.98，均差小于0.63g/m3，标准差小于1.22 g/m3。并利用LAPS系统建立新的目标函数，成功融合了BDS-GPS三维水汽观测，形成初始场并参与中尺度模式预报，分析显示与PWV比较，三维层析水汽密度的加入更能提高降水，尤其是暴雨以上量级的降水预报准确率。.通过项目的实施，解决了辅助大气观测参与三维水汽层析、北斗-GPS斜路径总湿延迟算法问题、层析解算的大气湿折射度不能直接应用、LAPS系统融合北斗-GPS三维水汽密度参数等问题，建立了一套基于观测约束的三维水汽层析新算法，形成了完整的基于北斗导航定位系统的水汽观测技术，为北斗-GPS三维水汽资料投入气象业务应用提供了支撑。