FY-4A GIIRS高时间分辨率温湿度廓线反演及其在台风中的应用

针对2020年第9号台风"美莎克"期间FY-4A 高光谱红外干涉式大气垂直探测仪GIIRS每15 min一次的目标区跟踪加密观测资料,用三维卷积神经网络算法反演的全天空大气温度、湿度廓线分析了台风处于生命史不同阶段时暖心结构和湿度场结构的演变特征。结果表明:卷积神经网络的深度机器学习算法可以用来反演全天空的三维大气温度和湿度垂直廓线,不光适用范围广(晴空和有云视场)、反演精度高,而且反演速度快。利用静止卫星平台高时间分辨率的特性,反演得到的温度、湿度廓线可以细致追踪台风处于发展、成熟和登陆等阶段时暖心结构和湿度场的时空演变特征。台风从发展阶段(热带风暴和强热带风暴)到成熟阶段至登陆消亡时,暖心首先出现在对流层中高层较薄的区域,随着台风强度的加强,深厚的暖心结构明显、强度增加,水平面积增大且垂直往下延伸。由于对流云中强上升气流的输送水汽正距平区逐渐上传至300 hPa,台风最强时密闭云区与四周下沉气流区比湿差高到8 K·kg^-1。暖心结构和高湿度中心随着台风登陆而逐渐消失。

Temperature and humidity retrieval of FY-4A GIIRS with high temporal resolution and its application in typhoon

Based on FY-4A GIIRS (Geostationary Interferometric Infrared Sounder) tracking observation data to the target area with high-time density (every 15 minutes) during typhoon No. 2009, the three-dimensional convolution neural network algorithm was adopted to retrieve the atmospheric temperature and humidity profiles under all-sky condition in this paper. The GIIRS observed brightness temperatures and the retrievals of atmospheric temperature and humidity profile were used to analyze the evolution characteristics of typhoon warm core and humidity field structure during different stages in typhoon life history. Results show that the convolution neural network as deep machine learning algorithm is effective for retrieving three-dimensional atmospheric temperature and humidity profile under all-sky condition, not only suitable for clear sky and cloud field of view but with high accuracy and high speed. Utilizing the high temporal resolution characteristics of the geostationary satellite platform, the temperature and humidity profile retrievals can track the temporal and spatial evolution of typhoon warm core and humidity field structure in detail during different stages. The warm core first appears in the thin region of the middle and upper troposphere. The deep warm core structure is obvious, the intensity increases, the horizontal area increases and extends vertically downward with the strengthening of typhoon intensity from development stage (tropical storm and severe tropical storm), maturity (typhoon) to landfall and dissipation stages. Due to the transport of strong updraft in convective cloud, the positive water vapor anomaly gradually uploads to 300 hPa height. When the typhoon is strongest, the humidity difference between the center closed cloud area and the surrounding downdraft area is up to 8 K·kg^-1. Warm core structure and high humidity center gradually disappear with the typhoon landing.