GRAPES西南低涡集合预报评估及ETKF初值扰动方法改进

发布时间：2018-04-11 21:12

本文选题：西南低涡 + 移动路径　；参考：《中国气象科学研究院》2017年硕士论文

【摘要】：西南低涡是对我国降水天气有重要影响的中小尺度天气系统。目前国内外对于西南涡的发生发展、三维结构、温湿结构等研究较多,但对于西南低涡预报路径研究甚少。我国已经建立基于GRAPES-Meso中尺度模式区域集合预报系统(GRAPES-REPS),然而目前缺乏一种适用于格点分析资料的西南低涡识别方法,也尚不清楚模式对西南低涡这类中尺度系统的集合预报效果,尚未针对西南低涡集合预报中的不确定性构造更为合理的扰动初值。为了系统地研究GRAPES区域集合预报对西南低涡的预报能力,本文首先设计了一种适应于高分辨率格点分析资料的西南低涡识别方法,并通过一系列的检验来验证该方法的合理性及有效性。利用该方法开展GRAPES区域集合预报(GRAPES-REPS)对西南低涡移动路径预报检验评估,并通过分析西南低涡强度、降水等方面的内容整体评估GRAPES-REPS对西南低涡的预报效果。由于西南低涡是浅薄的中低层天气系统,本文在ETKF初值扰动方案中增加观测空间垂直层次,以解决中低层观测信息不足的问题,并设计了五组投影矩阵观测变量组合对比试验,试图通过改进ETKF初值扰动方法来提高GRAPES区域集合预报对西南低涡的预报技巧。得到了以下主要结论:本文所设计的识别方法(HVW识别方法)能够有效识别出高精度格点资料中的西南低涡过程,与格点实况的吻合率达87.5%,并减少了格点实况主观分析的空报,与天气图实况对比,该方法命中率接近80%,充分说明了该方法能够有效捕捉西南低涡,验证了西南低涡HVW识别方法技术的有效性与合理性。随着数值预报逐步转向精细化预报,海量数据涌现,该方法的应用可以极大地减轻预报员的分析工作量。对西南低涡集合预报的检验发现GRAPES模式对西南低涡预报的命中率较高,但四川盆地九龙地区空报较多。控制预报和大部分集合成员对低涡生成时间预报偏早,对消亡预报偏晚。对西南低涡移动路径在24h预报时效内较合理,24h预报时效之后存在不同程度的偏北,集合预报平均的移动路径误差明显小于控制预报,体现出明显优势。对西南低涡中心涡度值预报偏大,位势高度值预报偏低。东移型西南低涡移动路径偏北,导致降水落区偏北,而停滞型西南低涡降水则表现出略偏西的特征。对ETKF初值扰动方案改进试验的分析结果表明,ETKF初值扰动方案中增加观测空间垂直层次能够显著提高预报效果。此外,五组观测变量选择对比试验的分析结果发现,随着参与计算的变量的增加,α达到稳定所需的时间更长,放大因子的最终状态更小。对变量在整个模拟区域的均方根误差和离散度的分析发现,Test-1(U、V)的组合是最优的,Test-2是次优的方案。对一次典型的西南低涡过程的路径预报分析发现,增加投影矩阵的观测变量使西南低涡中心的预报更加集中,径预报的误差也进一步减小;对降水的分析发现,增加观测空间垂直层次试验(Test-1)对降雨落区偏北有所改善;增加投影矩阵观测变量会使贵州中部的降水概率值有进一步提高,然而也会导致强降水概率预报大值区偏北。结合路径预报和降水落区预报两方面来看,Test-5对路径预报误差最小,对降水概率预报值最大;Test-1、2路径误差略逊于Test-5,对降水落区预报优于Test-5。
[Abstract]:The southwest vortex is small and medium scale weather system has an important impact on China's precipitation. At home and abroad for the development of the southwest vortex, the three-dimensional structure of the temperature and humidity structure, but the study on the prediction of vortex path. Little Southwest China has established a scale model of regional GRAPES-Meso ensemble prediction system based on (GRAPES-REPS), however the current lack of a recognition method for the southwest vortex lattice analysis data, it is not clear the ensemble model of this kind of scale system of Southwest Vortex in the southwest vortex, has yet to set forecast uncertainty in the structure is more reasonable. The initial disturbance in order to study the GRAPES ensemble prediction ability of southwest region the vortex, we designed a suit of Southwest Vortex identification method in high resolution lattice analysis data, and through a series of tests to check This method is reasonable and effective. To carry out the GRAPES region by using the method of ensemble prediction (GRAPES-REPS) test on the southwest vortex moving path prediction, and through the analysis of the southwest vortex intensity, precipitation forecast and other aspects of the overall assessment of the GRAPES-REPS of the southwest vortex. Because of the southwest vortex is low layer shallow weather system, the disturbance the initial increase in ETKF observed vertical space level scheme, in order to solve the problem of low layer observation information, and observation variable combination were compared between the five groups of projection matrix design, tries to change into the ETKF initial perturbation method to improve the GRAPES regional ensemble forecast skill of southwest vortex. The main conclusions are as follows: the identification method the design (HVW recognition method) can effectively identify the southwest vortex lattice in the process of high precision data, lattice live with The rate of up to 87.5%, and reduce the empty newspaper lattice live subjective analysis, and comparison of the method of live weather map, the hit rate of nearly 80%, which indicates that this method can effectively capture the southwest vortex, verified the technology of southwest vortex HVW recognition method is effective and reasonable. With the numerical prediction gradually turned to the fine forecast of mass the emergence of data, this method can reduce the workload of the forecaster greatly. Inspection of the southwest vortex ensemble found GRAPES mode on the southwest vortex prediction hit rate is higher, but the Kowloon area of Sichuan basin. The prediction and control of more space at most of the collection members of the vortex generation time forecast earlier, to forecast late death on the southwest vortex moving path in the 24h forecast period is reasonable, after the 24h forecast period there are different degrees of North, set the moving path is obviously less than the average forecast error The control prediction reflects the obvious advantage of the southwest vortex center. The vorticity value forecast is too large, the geopotential height prediction is low. The southwest vortex moving eastward moving path leading to precipitation area north, north, and southwest vortex stagnation type precipitation is characterized by somewhat to the West. The analysis results of ETKF initial perturbation scheme to improve the test show that the ETKF initial perturbation scheme in observation space increased vertical level can significantly improve the prediction effect. In addition, five groups of observation and analysis of variable selection results of comparison tests found that with the increase in the calculation of the variables, alpha stable reach required a longer period of time, the final state is smaller. The amplification factor variables in the root mean square error of the whole the simulation area and dispersion analysis, Test-1 (U, V) of the combination is the best, Test-2 is suboptimal. The path prediction of a typical southwest vortex process analysis, Increase the projection matrix of the observed variables to the southwest vortex center forecast is more concentrated, size of the forecasting error is further reduced; found on precipitation analysis, increase the observation space level test (Test-1) on the vertical rainfall in North has improved; increase the projection matrix of observation variables will make the probability of precipitation in central Guizhou have further improved however, can also lead to strong precipitation probability forecast value of large areas of North. Combining the path prediction and precipitation area forecast of two in terms of Test-5 on track prediction error, the maximum value of precipitation probability forecast; Test-1,2 path error in Test-5 Chalvexun, for rainfall forecast is better than that of Test-5.

【学位授予单位】：中国气象科学研究院
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P456.7

【参考文献】