GPT3模型中国区域大气剖面应用精度分析

Analysis of application accuracy of GPT3 model in China’s regional atmospheric profile

对流层延迟是卫星导航定位最主要的误差来源之一,精确计算对流层延迟有助于模糊度的收敛及定位精度的提高。目前应用最广、精度最高的对流层经验模型是全球气压气温模型,为了验证GPT3模型计算中国区域地表至11 km大气剖面对流层延迟改正的精度,利用2011—2020年中国区域82个参与全球气象交换的测站的无线电探空数据,对GPT3的气压(P)、气温(T)、水汽压(E)以及加权平均温度(Tm)进行精度检验及分析,实验结果表明,GPT3模型精度受纬度和高程影响较大,其中GPT3-P和GPT3-Tm受纬度影响显著;GPT3-P在地表RMS为8.02 hPa,而在地表至11 km其RMS为20.01 hPa,说明模型地表精度要优于大气剖面精度,GPT3-Tm呈现同样的规律,而GPT3-T的地表以及地表至11 km的RMS分别为7.94 K、7.53 K,GPT3-E的RMS分别为2.42 hPa、1.97 hPa;模型在不同年积日的精度存在差异,呈现一定的季节特性,但其精度在长时间区间内没有明显变化。总体而言,GPT3模型在中国地区范围内具有较高的精度和稳定性,但需要进一步优化其高程改正方法。

Tropospheric delay can result in errors in satellite navigation and positioning.Accurate calculation of tropospheric delay helps to converge the ambiguity and improve the positioning accuracy.At present, the most accurate tropospheric empirical model is the Global Pressure and Temperature 3(GPT3).In order to verify the accuracy and reliability of the GPT3 model in calculating the tropospheric delay correction from the surface to 11 km of the atmospheric profile in China, this paper uses radiosonde data from 82 stations participating in the global weather exchange in China from 2011 to 2020 to analyze the accuracy of air pressure(P),temperature(T),air pressure(E) and weighted mean temperature(Tm) of GPT3.The experimental results are as follows: the accuracy of GPT3 model is greatly affected by latitude and elevation.Among them, GPT3-P and GPT3-Tmare significantly affected by latitude;GPT3-P has an RMS of 8.02 hPa at the surface, and an RMS of 20.01 hPa from the surface to 11 km, indicating that the model’s surface accuracy is higher than the atmospheric profile accuracy, and the result is also applicable to the GPT3-Tm.The surface RMS of the GPT3-T and the surface to 11 km RMS are 7.94 K and 7.53 K,and those of GPT3-E are respectively 2.42 hPa and 1.97 hPa, and the accuracy of the model changes with the season.In general, the GPT3 model has high accuracy and stability in China, but its elevation correction method needs to be further optimized.