顾及衰减系数时空变化的全球PWV垂直改正模型

A global vertical correction model of PWV considering the spatial-temporal variation of decay coefficient

大气可降水量(precipitable water vapor,PWV)在多尺度气候变化及大气物理过程中扮演着重要的角色。为提高PWV垂直改正精度进一步扩展各类PWV产品的空间应用性,本文基于2010—2019年ERA5再分析资料构建了一种顾及PWV垂直衰减系数时空变化的全球适用范围的分层格网模型(GPWVCS)。同时联合2020年ERA5及无线电探空PWV,评估了本文模型在全球范围内的精度及适用性。结果表明,相比于经验模型及未分层的GPWVC模型,GPWVCS模型有效提升了PWV的垂直改正精度。以ERA5 PWV为参考,全球范围内GPWVCS模型修正PWV的RMS不超过1.9 mm。以探空数据为参考,GPWVCS模型在热带、温带、寒带及全球范围的年均RMS分别为2.24、1.29、0.44、1.44 mm,较经验模型分别提升34.6%、14.1%、10.9%及21.4%,较GPWVC模型分别提升6.4%、5.8%、9.4%及6.0%。GPWVCS的分层算法最大限度地削弱了PWV指数外推的误差累积影响,本文开发的水平分辨率为1°×1°、2°×2°及5°×5°的模型均能够显著提升全球范围内多种高差下PWV的垂直改正效果,用户可以根据计算效率及精度需求自行选择最佳模型。

Precipitable water vapor(PWV)plays an important role in multi-scale climate change and atmospheric physical processes.To improve the accuracy of vertical correction of PWV and further extend the spatial applications of multi-source PWV products,we propose a global stratification grid model named as GPWVCS based on ERA5 reanalysis data from 2010 to 2019,considering the spatial-temporal variation of PWV vertical decay coefficient.And the accuracy and application of the new model on a global scale are evaluated using ERA5 and radiosonde derived-PWV.The results show that compared with the empirical model and the unstratified GPWVC model,the GPWVCS model effectively improves the vertical correction accuracy of PWV.Taking ERA5 PWV as reference,the RMS of PWV adjusted by GPWVCS model in each region over the globe is less than 1.9 mm.Taking radiosonde data profiles as reference,the annual mean RMS values of GPWVCS model in the tropical,temperate,frigid zone and globe are 2.24,1.29,0.44 and 1.44 mm,respectively,which achieve accuracy improvements of 34.6%,14.1%,10.9%and 21.4%compared with the empirical model,and they are corresponding to 6.4%,5.8%,9.4%and 6.0%improvement against the GPWVC model.The stratification algorithm of GPWVCS model greatly weakens the impact of the PWV accumulated error which is attributed to the exponential extrapolation.All three models with horizontal-resolutions of 1°×1°,2°×2°and 5°×5°developed in this study can significantly improve the global performance of vertical correction for PWV in terms of different height differences,and users can choose the best model according to the requirements of computational efficiency and accuracy.