摘要
以中国气象科学研究院灾害天气国家重点实验室的区域雷达组网三维数字产品为基础,采用区域增长法实现二维风暴的识别,进而在垂直方向上进行关联以实现三维风暴的识别,并提取三维风暴的特征参数对相邻时刻的风暴进行跟踪,在此基础上对风暴未来时刻的位置及大小进行预报。个例分析结果表明:(1)用区域增长法及垂直方向的投影关联法能够实现三维风暴的自动识别,从风暴特征参数随时间的变化可以看出,体积的变化是风暴不同发展阶段的主要特征,而风暴的平均反射率因子及最大反射率因子随时间的变化特征不明显;(2)采用最优法对相邻时刻的风暴进行匹配,并考虑了合并与分裂的情况,与观测资料的对比可以看出,跟踪还是比较合理,在风暴发生合并或分裂的情况下也能够正确跟踪。对风暴历史时刻的位置通过最小二乘拟合法得到的风暴移向、移速随时间的变化曲线比较光滑,没有出现跳变的现象。(3)风暴位置的预报要好于风暴大小及VIL的预报,其主要原因是由于风暴在不断地发展变化,其合并、分裂也是经常发生的,对于未来发展趋势的预报,仅根据风暴历史时刻的发展趋势还是很难预报的。总体来说,预报误差随预报时效的增大而增加。
Based on the radar mosaic 3D data with clutter elimination algorithm and an extensive quality control program developed by the State Key Laboratory of Severe Weather,Chinese Academy of Meteorological Sciences,a methodology is presented for the automated identification,tracking and nowcasting of storm.The convective storms of each plane are detected using the region-growing method based on constant altitude surfaces（CAPPI） between 0.5 km and 16 km of radar mosaic,which are associated in the vertical direction in order to get whole storm.The characteristic parameters of whole storm such as storm center,top,base,VIL,volume,moving vector etc.are then computed.The detected storms are tracked in successive images based on combinatorial optimization method,with some geometric logic to deal with merges and splits.Then forecasts are made for the several parameters such as storm center,projected-area ellipse and VIL.The performance of the detection and nowcasting are evaluated by applied in several convective cases.It shows that the identification method could achieve automated detect storm,the volume change is the crisis of main features of different stages of development from the storm characteristic parameters change,while the storm,the average reflectivity factor and maximum reflectivity factor variation with time is not obvious.It is considered that the tracking is relatively reasonable even in the case of merge and split by analyzing the observational radar data.The forecast position of storm is better than the size and VIL,mainly due to the storm development and change constantly,as well as storm merge,split occurred frequently.It is very difficult to forecast the future development trend only based on its history.Overall,the prediction error increases with the forecast period increasing.
出处
《高原气象》
CSCD
北大核心
2010年第6期1546-1555,共10页
Plateau Meteorology
基金
中国气象科学研究院基本科研业务费项目(2007Y004)
科技部公益性行业科研专项(GYHY200906003)
福建省科技厅重点项目(2008Y0027)共同资助
关键词
雷达拼图
自动识别
分裂与合并
临近预报
Radar mosaic
Automated identification
Merge and split
Nowcasting
