Retrieval of atmospheric and oceanic parameters and the relevant numerical calculation

It is well known that retrieval of parameters is usually ill-posed and highly nonlinear, so parameter retrieval problems are very difficult. There are still many important theoretical issues under research, although great success has been achieved in data assimilation in meteorology and oceanography. This paper reviews the recent research on parameter retrieval, especially that of the authors. First, some concepts and issues of parameter retrieval are introduced and the state-of-the-art parameter retrieval technology in meteorology and oceanography is reviewed briefly, and then atmospheric and oceanic parameters are retrieved using the variational data assimilation method combined with the regularization techniques in four examples： retrieval of the vertical eddy diffusion coefficient; of the turbulivity of the atmospheric boundary layer; of wind from Doppler radar data, and of the physical process parameters. Model parameter retrieval with global and local observations is also introduced.     

利用改进的GoogLeNet深度学习模型识别COSMIC-2掩星信号中的反射信号

GPS(Global Positioning System)掩星探测技术作为一种先进的大气探测手段,已广泛用于数值天气预报、气候和空间天气研究。掩星探测存在的问题之一是容易受到地球表面反射信号的干扰,识别和分离掩星探测信号中的反射信号有助于将掩星数据同化到数值天气预报系统中去,具有重要意义。本文提出一种基于改进的GoogLeNet深度学习模型（Im-GNet）,并应用于COSMIC-2掩星探测数据来识别反射信号。本文选择了2020年1月1～9日的COSMIC-2掩星数据（conPhs文件）,进行质量控制后,利用无线电全息方法得到掩星信号的无线电全息功率谱密度图像,然后训练得到Im-GNet深度学习模型,Im-GNet模型测试的准确率达到了96.4%,显著高于支持向量机（SVM）方法的结果。本文还分析了反射信号对掩星数据的影响,掩星事件的地理分布以及掩星反演数据（atmPrf文件）与NCEP再分析资料的12 h预报值（avnPrf文件）的折射率比较表明：有反射信号的掩星事件数据质量更好,所包含的大气信息更丰富。     

Can the uncertainties of Madden-Jullian Oscillation cause a significant “Spring Predictability Barrier” for ENSO events?

With the Zebiak-Cane model and a parameterized stochastic representation of intraseasonal forcing, the impact of the uncertainties of Madden-Jullian Oscillation (MJO) on the ??Spring Predictability Barrier (SPB)?? for El Ni?o-Southern Oscillation (ENSO) prediction is studied. The parameterized form of MJO forcing is added physically to the Zebiak-Cane model to obtain the so-called Zebiak-Cane-MJO model and then the effects of initial error, stochastic model error, and their joint error mode on the SPB associated with El Ni?o prediction are estimated. The results show that the model errors caused by stochastic MJO forcing could hardly lead to a significant SPB while initial errors can do; furthermore, the joint error mode of initial error and model error associated with the stochastic MJO forcing can also lead to a significant SPB. These demonstrate that the initial error is probably the main error source of the SPB, which may provide a theoretical foundation of data assimilation for ENSO forecasts.     

OSG中三维场景构建的关键技术

随着基于OpenGL的各种三维渲染引擎的日趋成熟,如何更好地利用这些引擎已成为人们首先要解决的问题,而OpenSceneGraph作为一款高性能的、开源的渲染引擎也越来越受到人们的关注。系统地对OSG的结构和特性进行了分析,并对OSG中三维场景的构建过程以及其中的一些关键技术如三维场景的数据组织、场景的裁剪技术、渲染的流程等进行了研究。     

基于正交匹配小波算法的台风云系提取研究

西北太平洋是世界上台风发生频率最高、强度最强的一个海区,因此对西北太平洋台风的发生发展进行相关的分析研究是很有意义的。论文中利用静止卫星观测数据,进行了滤波、去噪等预处理,及综合优化的云检测处理;在此基础上,利用改进的正交匹配小波算法,实现了西北太平洋地区台风云系特征的提取。运用该算法提取出的台风云系特征参数简单,与其他算法比较,尽管提取精度有待改进,但是用于台风的监测预警可提高效率,适用于台风预报预警监测的业务化应用。