期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
深度学习模型TAGAN在强对流回波临近预报中的应用 被引量:2
1
作者 胡家晖 卢楚翰 +1 位作者 姜有山 何婧 《大气科学》 CSCD 北大核心 2022年第4期805-818,共14页
近年来深度学习模型在解决对防灾减灾影响巨大且极具挑战性的临近预报问题的应用中日益增多。本文中,我们把临近预报作为一个时空序列预测的任务,将雷达反射率因子作为试验对象,使用基于对抗神经网络(GAN)优化构建的TAGAN深度学习模型... 近年来深度学习模型在解决对防灾减灾影响巨大且极具挑战性的临近预报问题的应用中日益增多。本文中,我们把临近预报作为一个时空序列预测的任务,将雷达反射率因子作为试验对象,使用基于对抗神经网络(GAN)优化构建的TAGAN深度学习模型预测未来1小时的雷达回波图像,并且与Rover光流法、基于卷积神经网络的3D U-Net模型进行对比试验。选取2018年全球气象AI挑战赛雷达回波数据集进行训练与测试,检验结果表明TAGAN模型在命中率(POD),虚警率(FAR),临界成功指数(CSI)以及相关系数等多种评分上要优于传统的光流法和对比的3D U-Net深度学习模型,TAGAN模型在以上的检验评分表现出色,并且随预测时间的增加较之传统光流模型效果更优,这为拓展和提升深度学习模型在临近天气预报中的应用提供了参考依据。 展开更多
关键词 临近预报 时空预测 深度学习 雷达回波
下载PDF
Evaluation of the CAM and PX Surface Layer Parameterization Schemes for Momentum and Sensible Heat Fluxes Using Observations
2
作者 youshan jiang Dongqing LIU Gang LIU 《Journal of Meteorological Research》 SCIE CSCD 2018年第6期1026-1040,共15页
In this study,the performances of the Community Atmosphere Model(CAM)and Pleim–Xiu(PX)surface layer parameterization schemes are investigated by using field observations.The parameterization schemes are evaluated aga... In this study,the performances of the Community Atmosphere Model(CAM)and Pleim–Xiu(PX)surface layer parameterization schemes are investigated by using field observations.The parameterization schemes are evaluated against continuous momentum and sensible heat flux observations measured at two flat and homogeneous grassland sites in the suburb of Nanjing,eastern China.The observations were conducted from 30 December 2014 to 18 April 2017 at Jiangxinzhou and from 9 February 2015 to 26 March 2018 at Jiangning.It is found that the momentum flux is overall in good agreement with the observation,and the sensible heat flux is overestimated.The parameterizations of the momentum and sensible heat fluxes well capture the diurnal and seasonal patterns seen in the observations at the two sites.At Jiangxinzhou,the PX parameterization underestimates the momentum flux throughout the day and the CAM parameterization slightly overestimates it around the noon,while they underestimate the momentum flux throughout the year.The two parameterizations overestimate the sensible heat flux in the daytime as well as over the entire year.At Jiangning,the two parameterizations overestimate the momentum flux throughout the day and the sensible heat flux in the daytime,and overestimate both of them over the entire year.The two parameterizations are not significantly different from each other in reproducing the turbulent fluxes at the same site,while they perform differently at the two sites in terms of statistics.In addition,the parameterized fluxes increase with increased roughness length. 展开更多
关键词 EVALUATION PARAMETERIZATION turbulent fluxes surface layer
原文传递
上一页 1 下一页 到第
使用帮助 返回顶部