GPT2w模型反演大气可降水量的研究

Study on Retrieving Precipitable Water Vapor by GPT2w Model

利用地基实测气象资料分析了GPT2w(global pressure and temprature 2 wet)模型估算的气象数据的精度,并将GPT2w估算的气象要素结合地基GNSS测站观测资料进行了大气可降水量(precipitable water vapor,PWV)的反演,结果表明:就BJFS测站在夏秋之季而言,无论是日尺度还是小时尺度上,由GPT2w估算的气象要素来反演的PWV与利用地基实测气象要素来反演的PWV的均方根(root mean square,RMS)优于2 mm,且两种尺度上两者RMS的偏差为亚毫米级,这为地基GNSS测站气象数据缺失时反演PWV提供了一种参考思路。

The accuracy of meteorological data estimated by GPT2 w model was evaluated by using ground-measured meteorological data.Then the precipitable water vapor was retrieved by meteorological elements estimated by GPT2 w and ground-based GNSS station observation data.The results show that,As far as the BJFS station is in the summer and autumn season,the RMS of the PWV derived from GPT2 w and the PWV derived from ground-based measured meteorological elements are better than2 mm in both daily and hourly scales.The deviation of the RMS of between the two scales is sub-millimeter.This provides a reference idea to invert PWV when the meteorological data of the ground-based GNSS station is missing.