【作者】 韦林祯；

【导师】 王萍；

【作者基本信息】 天津大学 ， 模式识别与智能系统， 2010， 硕士

【摘要】 边界层辐合线具有很强的气象风暴灾害的预示能力,它在多普勒雷达图像上表现为一条蜿蜒的、可能有岔路的、宽度不定的、取值变化的、若隐若现的带状区域。本文对“探测窗”协同加深“印象”算法进行改进,并对辐合线相关潜势特征进行具体提取、统计与分析,用于识别这种具有复杂背景的非规则带状区域,获得了更好的处理结果。本文主要完成了以下两部分工作:一、改进辐合线的识别算法并对辐合线区域提取骨架。本文针对“探测窗”协同加深“印象”算法原理,对识别问题进行总结,根据大量样本实验,对阈值与相关参数进行调整,再通过细化得到辐合线骨架。二、利用剖面图知识,对辐合线区域及其附近区域的水汽分布特征进行分析总结;利用对辐合线骨架两侧的风场速度信息提取,分析辐合线的形成特征、运动特征,以及与未来出现的风暴天气的联系,给出一定的结论。综上所述,本文为边界层辐合线的识别与风暴预测提供了新的思路,所涉及算法均已编程实现,并在样本测试过程中表现出较强的稳定性和令人满意的测试效果。更多还原

【Abstract】 Boundary convergent lines have strong ability to predict weather storms, which are shown as some winding, possibly to have the branch road and looming belt-shaped regions with uncertain width and change value in the Doppler radar images.This paper makes improvements to the algorithm including "detection window" and deepen "impression", and identifies potential features of the boundary convergent lines and casts statistical analysis, applies the above algorithm to the complex the irregular belt-shaped region and achieves satisfactory results. This paper has mainly completed the following two tasks:First, improves the algorithm of identifying boundary convergent lines and extracts skeleton of region of boundary convergent lines. This paper studies the principle of the algorithm including "detection window" and deepen "impression", summarizes problem identification methodology, based on large size of sample survey modifies related threshold value and parameters, and gets skeleton of boundary convergent lines through specification.Second, applies sectional drawing theory to analyze and summarize water and steam distribution characteristics within and near region of boundary convergent lines; uses wind field speed on both sides of skeleton of boundary convergent lines to extract information, analyze characteristics of boundary convergent lines convergence and movements as well as the relationship between these characteristics and potential stormy weather and provide related conclusions.In sum, it has developed a new method for boundary convergent lines identification and storm prediction, realized related algorithms and its coding, and achieved strong robustness during sample tests and other satisfactory testing results.更多还原