摘要
针对在利用全球卫星导航系统(GNSS)反演高精度的大气水汽产品时,容易缺少实测气象参数的问题,提出利用全球气温气压3(GPT3)模型计算的气温、气压参数,结合GPT3、贝维斯(Bevis)回归和改进Bevis回归计算的加权平均温度,来反演大气可降水量(PWV)的方法。3种基于GPT3模型计算PWV的方法分别记为G/PWV、GB/PWV和GBG/PWV。利用美国怀俄明州立大学提供的探空站实测PWV数据作为真值,对计算的PWV进行精度评估。实验结果表明:G/PWV、GB/PWV和GBG/PWV均能够达到毫米级精度,它们的年均偏差分别为0.06、0.17、0.42 mm,均方根误差分别为4.47、4.48、4.54 mm;对于不同的测站和地理分区,3种方法反演出的PWV精度有所不同且呈现出不同的规律;整体而言,在中国陆地区域内利用GPT3模型反演PWV,能够达到较高的精度和稳定性。
In view of the problem that it is easy to lack measured meteorological parameters when using Global Navigation Satellite System(GNSS)to inverse high-precision atmospheric moisture products,this paper proposed the Precipitable Water Vapor(PWV)inversion methods by the weighted mean temperature calculated by Global Pressure and Temperature 3(GPT3)model,Bevis regression and improved Bevis regression combined with the temperature and pressure parameters calculated by the GPT3.Three methods based on the GPT3 model to calculate PWV were recorded as G/PWV,GB/PWV and GBG/PWV,respectively.The measured PWV data of the sounding station provided by Wyoming State University as the true value was used to evaluate the accuracy of the calculated PWV.The experimental results showed that G/PWV,GB/PWV and GBG/PWV could achieve millimeter accuracy.Their average annual deviation were 0.06,0.17 and 0.42 mm,respectively,and the average square root error were 4.47,4.48 and 4.54 mm.For different stations and geographical regions,the PWV accuracy of the three methods was different and presented different laws.On the whole,the inversion of PWV using the GPT3 model in the land area of China could achieve high accuracy and stability.
作者
陈发源
王新志
金晓毅
CHEN Fayuan;WANG Xinzhi;JIN Xiaoyi(School of Remote Sensing&Geomatics Engineering,Nanjing University of Information Science&Technology,Nanjing 210044,China)
出处
《导航定位学报》
CSCD
2021年第6期38-44,共7页
Journal of Navigation and Positioning
基金
国家自然科学基金项目(12073012)
中科院先导A专项课题项目(XDA23040100)。
关键词
全球气温气压3模型
大气可降水量
加权平均温度
精度评定
global pressure and temperature 3 model
precipitable water vapor
weighted mean temperature
accuracy evaluation
