江西赣州地区地基GNSS反演水汽转换因子模型建立

Establishment of Conversion Factor Model of Water Vapor Inversion by Ground-based GNSS in Ganzhou of Jiangxi Province

由于赣州地区夏季降水量较大,易发生洪涝灾害,高精度反演大气可降水量(Precipitable Water Vapor,PWV)可实时监控水汽变化,预防洪涝灾害的发生。为提高赣州地区地基GNSS反演大气可降水量的精度,采用赣州探空站2016年~2018年全年的数据,通过数值积分的方式获得测站点的大气加权平均温度(T_(m)),根据T_(m)与地表温度(T_(s))、地表水汽压(E_(s))的变化规律,建立了基于T_(s)一元线性回归模型、基于T_(s)与E_(s)的二元回归模型。以2019年探空站数据作为预测数据,结果表明:赣州地区本地化模型T_(m)2精度最高,拟合RMSE为2.34K,预测RMSE为2.39 K,较Bevis模型提升了41.19%,较李建国模型提升了43.44%。本地化模型能较好地消除Bevis模型和李建国模型在赣州区域的系统偏差,提高了天顶湿延迟ZWD转换大气可降水量的精度。

Due to the large precipitation in summer,Ganzhou area is prone to flood disaster.High precision inversion of atmospheric precipitable water vapor(PWV)can monitor water vapor changes in real time,thus can prevent the occurrence of flood and waterlogging disasters.In order to improve the accuracy of ground-based GNSS inversion of PWV in Ganzhou area,the weighted mean temperature of the atmosphere(T_(m))of the station is obtained by numerical integration using the data of Ganzhou radiosonde station from the year 2016 to 2018.According to the variation character of T_(m),surface temperature(T_(s))and surface water vapor pressure(E_(s)),a linear regression model based on T_(s)and a binary regression model based on T_(s)and E_(s)are established.The radiosonde station data in 2019 is used as the prediction data.The results indicate that the localized model T_(m)2 in Ganzhou has the highest accuracy with a fitted RMSE of 2.34K and a predicted RMSE of 2.39K,which is 41.19%higher than the Bevis model and 43.44%higher than the Li Jianguo model.The localized model can effectively eliminate the systematic bias of the Bevis model and Li Jianguo model in Ganzhou region,and improve the accuracy of zenith wet delay(ZWD)conversion to precipitable watervapor(PWV).