| 顾及气象数据的中国区域对流层延迟RBF神经网络优化模型 |
| |
| 引用本文： | 徐天河,李耸,王帅民,江楠. 顾及气象数据的中国区域对流层延迟RBF神经网络优化模型[J]. 测绘学报, 2022, 51(8): 1690-1707. DOI: 10.11947/j.AGCS.2022.20210480 |
| |
| 作者姓名： | 徐天河 李耸 王帅民 江楠 |
| |
| 作者单位： | 山东大学空间科学研究院,山东 威海 264209;河北工程大学矿业与测绘工程学院,河北邯郸 056038 |
| |
| 基金项目： | 国家重点研发计划(2020YFB0505800;2020YFB0505802);国家自然科学基金(41874032);山东省自然科学基金(ZR2020QD046;ZR2020MD045) |
| |
| 摘 要： | 本文基于单层气象数据(ERA5单层数据、实测气象参数)和多层气象数据(ERA5气压层数据、COSMIC掩星数据),分别采取模型法和积分法获取了我国236个陆态网GNSS测站的ZTD值,即ERA5S_ZTD、MET_ZTD、ERA5P_ZTD、RO_ZTD。以GNSS_ZTD为参考,按月评估了上述4种ZTD估计值的精度,结果表明:4种ZTD估计值的月平均RMSE依次为42.8、53.6、16.1和62.3 mm,其中基于积分法估计的ERA5P_ZTD精度最高,采用模型法计算的ERA5S_ZTD和MET_ZTD次之,而利用积分法获取的RO_ZTD值精度较低。为进一步提升利用气象数据估计ZTD值的精度,本文提出了基于RBF神经网络的对流层延迟改进模型。计算结果表明:改进模型获得的4种ZTD值与GNSS_ZTD之间的月RMSE平均值分别为23.5、32.1、14.2和40.8 mm,精度较原有ZTD估计值提升43.4%,36.3%,10.0%和34.4%。整体而言,改进模型估计ZTD值精度提升效果明显,且提升率与测站分布的密集程度有关。
|
| 关 键 词： | GNSS RBF神经网络 ERA5 COSMIC 对流层延迟 |
| 收稿时间： | 2021-08-20 |
| 修稿时间： | 2022-02-25 |
Improved tropospheric delay model for China using RBF neural network and meteorological data |
| |
| XU Tianhe,LI Song,WANG Shuaimin,JIANG Nan. Improved tropospheric delay model for China using RBF neural network and meteorological data[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(8): 1690-1707. DOI: 10.11947/j.AGCS.2022.20210480 |
| |
| Authors: | XU Tianhe LI Song WANG Shuaimin JIANG Nan |
| |
| Affiliation: | 1. Institute of Space Science, Shandong University, Weihai 264209, China;2. College of Mining and Geomatics, Hebei University of Engineering, Handan 056038, China |
| |
| Abstract: | Single-level meteorological products (ERA5 single-level data and measured meteorological parameters) and multi-level meteorological products (ERA5 pressure level data and COSMIC data) are used to estimate the ZTD of 236 CMONOC stations,namely ERA5S_ZTD,MET_ZTD,ERA5P_ZTD and RO_ZTD,based on the model method and the integration method respectively.Four ZTD estimations are evaluated with the reference of GNSS_ZTD;the results show that the average monthly RMSE are 42.8,53.6,16.1 and 62.3 mm,respectively.The accuracy of ERA5P_ZTD estimated by the integral method is the highest,that of ERA5S_ZTD and MET_ZTD calculated by the model method are the next.Estimating RO_ZTD with the integral method has the lowest accuracy.In order to further improve the accuracy of ZTD estimations,improved model of tropospheric delay is proposed with the RBF neural network in this paper.The calculation results show that:the average monthly RMSE between the ZTD from four improved models and GNSS_ZTD are 23.5,32.1,14.2 and 40.8 mm,which are reduced 43.4%,36.3%,10.0% and 34.4% than raw ZTD estimations.The overall modified effect of the improved model is obvious,and the improvement rate is related to the density of station distribution. |
| |
| Keywords: | |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《测绘学报》浏览原始摘要信息 |
|
点击此处可从《测绘学报》下载全文 |
|
