meteorological parameters,GPT2 model,CORS,precipitable water vapor," />
气象参数,GPT2模型,CORS,大气可降水量,"/>
meteorological parameters,GPT2 model,CORS,precipitable water vapor,"/>
Accuracy Analysis on GPS Precipitable Water Vapor Inversion Using GPT2 Models
Abstract The meteorological parameters (air temperature and air pressure) estimated by GPT2 are analyzed and assessed using adjacent radio stations in the area of Shandong province. Further, GPT2 model is applied to SDCORS precipitable water vapor inversion. The research shows that the average deviation of air temperature and air pressure estimated by the GPT2 model are -1.61 ℃ and 0.53 Pa, the average standard deviation is 2.84 ℃ and 4.42 Pa, and the mean root mean square error is 3.27 ℃ and 4.49 Pa, respectively. Additionally, the mean deviation of SDCORS/PWV is 1.22 mm, the mean value is 3.05 mm, the mean square error is 3.46 mm, which is higher than that of GPT model, and the reliability is strong. For CORS stations without meteorological sensors, based on the GPT2 model to estimate the temperature and pressure, it is helpful to use the regional CORS to invert the precipitable water vapor, and to further effectively monitor and forecast the atmospheric precipitable water.
