基于纹理和区域特征的台风卫星云图分割方法

在利用GMS红外卫星云图进行无眼台风自动定位方法的研究中,台风云系的分割是处理中关键的一步,文章提出了一种基于纹理和区域特征的台风云系分割方法。首先利用图像的分形维数和灰度特征对台风云系中的密蔽云区进行有效的识别后,然后启动基于区域约束的区域生长计算得到台风云系。     

台风卫星云图分割方法研究

本文针对台风卫星云图的具体特点,采用一种利用迭代模型并结合台风云系面积特征的分割方法,对台风卫星云图进行分割,取得了满意的效果。     

台风卫星云图分割方法研究

本文针对台风卫星云图的具体特点,采用一种利用迭代模型并结合台风云系面积特征的分割方法,对台风卫星云图进行分割,取得了满意的效果。     

一种适合识别台风的纹理分析法

一、引言在所有灾害性天气中,台风恐怕是最有破坏力的灾害性天气之一。台风所带来的狂风巨浪和暴风雨给海上航行的船只和沿海人民带来严重的灾害。因此研究台风的监测与预报问题,已成为气象研究工作者一直在探索的一个重要研究课题。利用气象卫星发回的卫星云图,我们可以监视台风并发出台风顶报。采用计算机自动识别台风,不仅使人     

卫星云图有眼台风中心定位算法研究

本文分析和研究了单幅红外卫星云图台风定位问题,初步研究了有眼台风的模式识别,提出有眼区台风的中心定位算法.根据气象领域知识,眼区处在台风主体云系最大内切圆圆心附近,算法先提取出台风主体云系,应用数学形态学求取台风最大内切圆中心和半径,在此范围内寻找台风眼区.为了排除云洞和云缝的干扰,提出使用距离和灰度信息组成判别式求取台风眼区.实验证明具有较高的定位精度,能适合气象预报业务化的要求.     

基于纹理方向整体分布特征的台风自动识别方法

为实现卫星云图上台风的自动识别，提出了一种基于纹理方向整体分布特征的台风云系图象自动识别方法，通过引入矢量矩的概念来表现图象纹理整体分布规律，该识别方法采用全局搜索方式，将一窗口在整幅图象上滑动，首先计算出窗口图象内各点的纹理方向，进而得出窗口图像的矢量矩，将矢量矩与阈值比较来判整幅图象是否为台风云系图像，实验结果表明，该方法能够识别不同类型和不同发展阶段的台风云系图象，能够很好地将台风云系与其他干扰云系区分开，具有较广泛的适应性和较高的识别率。     

静止卫星图像热带气旋云系自动识别

热带气旋对我国东南沿海地区国民经济和人民生命财产威胁巨大,静止卫星云图是热带气旋实时监测的主要数据源。热带气旋在卫星云图上的纹理特征与其它云系相似度高,为气旋云系的自动准确提取带来困难。本文在矢量矩概念的基础上,提出了旋转系数的概念来表征热带气旋的形态本质特征从而实现热带气旋的自动识别。建立了基于静止卫星图像,运用最大类间方差法确定目标云系分割阈值,结合云系面积和亮温分布特性,利用旋转系数进行热带气旋云系自动识别的方法流程。以1211台风海葵为例,在台风生成发展期、成熟期以及消亡期内,进行了改进前后方法识别率的对比实验,统计发现改进方法的识别率分别为76%、95%、78%,均高于原始方法的59%、90%、63%。实验表明改进方法分割的热带气旋云系更为完整,对各阶段的热带气旋云系识别率均更高。     

基于遗传算法初始群体优化的台风中心自动定位

针对误差梯度求导方法求解台风中心自动定位最优目标函数时,具有解的局部优化问题以及最优解获取困难的不足,本文将遗传算法应用到卫星云图台风中心自动定位的优化求解中,实现了台风中心的准确自动定位。同时,考虑到标准遗传操作中初始群体的随机生成对最优解的搜索具有很强的敏感性和不确定性问题,本文根据台风云系的灰度分布特征,对初始群体的生成进行了改进与优化,求得了最优数值解。多个台风中心定位仿真试验结果验证了该方法的合理性和可靠性。     

同化FY-3C/MWHTS观测资料反演的海面气压场对台风数值预报的影响

由中国风云三号C星（FY-3C）搭载的微波温湿探测仪（MWHTS）的亮温观测资料能够实时反演得到高分辨率、高精度的海面气压场。基于三维变分同化方法将FY-3C/MWHTS观测资料反演的海面气压场同化进入中尺度天气研究与预报（Weather Research and Forecasting, WRF）模式,以台风“Maria”和“Noru”为例,通过控制实验和同化试验的对比分析,探讨了同化反演的海面气压场对台风数值预报的影响。初始化敏感性试验结果表明,同化海面气压场使初始时刻台风中心气压与位置更接近实况,并且调整了台风初始温度场和风场的结构和分布。台风的数值预报结果表明:同化反演的海面气压场能够改进台风的路径和强度预报精度。     

基于云导风场的形成期台风定位

研究形成期台风云图的普遍性规律,根据云图对象的特殊性对求取运动对象光流场的传统MCC(最大相关系数)算法进行改进,由此生成的云导风矢量图能够准确体现风场绕台风中心旋转这一客观事实.本文将云导风矢量与台风中心的关系归纳为台风定位判定规则,编制的定位算法对于各类形成期台风具有普遍适应性,定位结果较令人满意.     

基于红外云图的台风中心智能定位方法

台风中心附近的云墙是同心圆状云带且其灰度值范围较固定,针对该特点,利用红外云图对台风中心进行智能定位,提出由云图预处理、Snake活动轮廓模型提取台风云墙轮廓点、最小二乘法拟合圆组成的3步定位方法。实验结果表明,该方法定位速度快、精度高,能满足气象业务的实时性要求。     

基于FY-3C MWHTS的台风降水反演算法研究

为估测台风带来的地表瞬时降雨率，利用FY-3C上搭载的微波湿温探测仪（Microwave Humidity and Temperature Sounder，MWHTS)的L1级在轨观测亮度温度数据与多卫星降水分析TMPA（Tropical Rainfall Measuring Mission（TRMM）Multi-Satellite Precipitation Analysis）3B42降水产品数据，通过多元线性回归和BP神经网络两种算法对台风区的降水情况进行了反演研究。结果表明，由这两种算法反演的降水分布图可以清晰地看到台风中心、云墙以及螺旋雨带等台风的位置、分布及结构信息，这与TMPA 3B42降水产品数据估测到的台风降水分布图相一致。此外，从定量的角度来看，TMPA 3B42降水数据与这两种反演算法反演的地表瞬时降水量（mm/hr）都具有较高的相关性和较小的偏差和均方根误差，反演的精度较高。故这两种算法都可以用来反演台风区的降水量，同时也表明FY-3C MWHTS微波在轨观测资料在台风区监测及降水研究中能发挥出较高的应用价值。     

Research on Typhoon Precipitation Retrieval Algorithm based on FY-3C MWHTS

In order to estimate the instantaneous precipitation rates brought by the typhoon, the Level 1 brightness temperatures from the Microwave Humidity and Temperature Sounder (MWHTS) onboard the FY-3C satellite and the Tropical Rainfall Measuring Mission (TRMM) Multi-Satellite Precipitation Analysis (TMPA) 3B42 precipitation product data are used to retrieve the precipitation rates in the typhoon area using the multiple linear regression and BP neural network retrieval algorithms. The results show that the precipitation distribution maps retrieved by these two algorithms can be clearly observed the location, distribution and structural information of the typhoons such as typhoon center, cloud wall and spiral rain belt, which are consistent with the TMPA 3B42 precipitation product data. In addition, from a quantitative point of view, the TMPA 3B42 precipitation data and surface precipitation rate (mm/hr) retrieved by these two precipitation retrieval algorithms reach higher correlation and smaller deviations and root mean square errors, and the retrieval accuracy is higher. Therefore, these two retrieval algorithms can be used to retrieve the precipitation in the typhoon area. It also shows that microwave on-orbit observation data from the FY-3C MWHTS can play a high application value in typhoon monitoring and precipitation research.     

Observations of typhoon eye using SAR and IR sensors

In this review, recent studies on the observations of typhoon eyes by images acquired by multiple sensors, including synthetic aperture radar (SAR), and infrared (IR) radiometer, are first summarized. Large horizontal distances between typhoon eyes on the ocean surface by SAR and those on the cloud top by IR sensors have been demonstrated; these have previously been ignored but should not be ignored in typhoon forecasts and numerical simulations. Then, based on nine published typhoon cases, the horizontal shifts and vertical tilt angles from the cloud-top typhoon eye locations by IR sensors on board the Feng-Yun 2 (FY-2) and Multi Functional Transport Satellite (MTSAT) to those at sea surface by SAR are further estimated. This shift difference between different sensors raises an issue on project distortion and navigation system errors for FY-2 and MTSAT satellites, which are of concern to both space agencies and data users. Finally, issues for current ongoing study and future research related to typhoon eyes are discussed, including rainband tracking between sensors for local wind speeds.     

基于不确定性理论的降水天气分类研究

首先介绍了云模型这种新兴的处理不确定性问题的工具,提出了一种基于云模型的层次聚类方法.该方法将原始数据集划分为许多小簇,然后分别用云模型来表示这些簇,再通过云综合的方法对这些小簇进行逐层合并,实现对数据集的聚类.将该方法应用于2006-2007年中国及周边亚太地区FY-2C卫星云图的聚类分析,得出不同降水类型的云图特征.最后以2006年7月6日FY-2C中国分区图所示地区对应的FY-2C卫星云图进行降水类型分类为实例,得到不同降水天气对应的云图辐射亮度值特征,并利用分类结果对实况云图进行降水天气的判别,结果表明该方法对大暴雨天气的判别效果良好.     

Research on Transfer Learning Methods for Classification of Typhoon Cloud Image

Aiming at the complexity of traditional methods for feature extraction about satellite cloud images, and the difficulty of developing deep convolutional neural network from scratch, a parameter-based transfer learning method for classifying typhoon intensity is proposed. Take typhoon satellite cloud images published by Japan Meteorological Agency, which includes 10 000 scenes among nearly 40 years to construct training and test typhoon datasets. Three deep convolutional neural networks, VGG16, InceptionV3 and ResNet50 are trained as source models on the large-scale ImageNet datasets. Considering the discrepancy between low-level features and high-level semantic features of typhoon cloud images, adapt the optimal number of transferable layers in neural networks and freeze weights of low-level network. Meanwhile, fine-tune surplus weights on typhoon dataset adaptively. Finally, a transferred prediction model which is suitable for small sample typhoon datasets, called T-typCNNs is proposed. Experimental results show that the T-typCNNs can achieve training accuracy of 95.081% and testing accuracy of 91.134%, 18.571% higher than using shallow convolutional neural network, 9.819% higher than training with source models from scratch.     

基于迁移学习及气象卫星云图的台风等级分类研究

针对传统卫星云图特征提取方法复杂且深度卷积神经网络（Deep Convolutional Neural Network, DCNN）模型开发困难的问题,提出一种基于参数迁移的台风等级分类方法。利用日本气象厅发布的近40 a 10 000多景台风云图数据,构建了适应于迁移学习的台风云图训练集和测试集。在大规模ImageNet源数据集上训练出3种源模型VGG16,InceptionV3和ResNet50,依据台风云图低层特征与高层语义特征的差异,适配网络最佳迁移层数并冻结低层权重,高层权重采用自适应微调策略,构建出了适用于台风小样本数据集的迁移预报模型T-typCNNs。实验结果表明:T-typCNNs模型在自建台风数据集上的训练精度为95.081%,验证精度可达91.134%,比利用浅层卷积神经网络训练出的精度高18.571%,相比于直接用源模型训练最多提高9.819%。     

Spatio-Temporal Data Mining for Typhoon Image Collection

Our research aims at discovering useful knowledge from the large collection of satellite images of typhoons using data mining approaches. We first introduce the creation of the typhoon image collection that consists of around 34,000 typhoon images for the northern and southern hemisphere, providing the medium-sized, richly-variational and quality-controlled data collection suitable for spatio-temporal data mining research. Next we apply several data mining approaches for this image collection. We start with spatial data mining, where principal component analysis is used for extracting basic components and reducing dimensionality, and it revealed that the major principal components describe latitudinal structures and spiral bands. Moreover, clustering procedures give the birds-eye-view visualization of typhoon cloud patterns. We then turn to temporal data mining, including state transition rules, but we demonstrate that it involves intrinsic difficulty associated with the nonlinear dynamics of the atmosphere, or chaos. Finally we briefly introduce our system IMET (Image Mining Environment for Typhoon analysis and prediction), which is designed for the intelligent and efficient searching and browsing of the typhoon image collection.