基于静止卫星红外云图的MCS自动识别与追踪

由于缺乏成熟的中尺度对流系统 (MCS) 自动追踪工具，使基于静止卫星红外云图的MCS研究的深度和广度受到制约。手工目测法识别追踪MCS，工作量大且容易造成误差，无法用于长时间序列卫星云图的MCS普查。该文结合我国中尺度对流天气的特点，在重新定义MCS云团识别标准的基础上，给出了一种基于图像处理和时间序列分析技术的MCS自动识别、存储、追踪和时间序列特征统计方法。通过对2003年淮河大水期间连续时次卫星云图的MCS自动识别、追踪和特征统计，对该方法的应用效果进行了检验，结果表明:该方法不仅能对每个时次的MCS进行快速识别，同时也能对多时次的MCS进行有效追踪。

MCS Identification and Tracking Based on Geo satellite IR Images

MCS (mesoscale convection systems) are significant weather systems causing heavy rain, hail and other severe weather events. Many disastrous weathers are usually caused by strong convection systems of 10—200 kilometers, but they are very difficult to forecast in operation. Geostationary satellite infrared imagery with higher spatial and temporal resolution provides much practical information for identifying and tracking MCS automatically from a broader perspective. Many researches are implemented on MCS based on geostationary satellite infrared imagery, amending the MCS judgment standard according to the actual condition of the weather. However, the lack of mature auto-tracking software has limited the extensive surveying of MCS using geostationary satellite. Artificial method is too onerous and error prone.An automatic method of identifying, saving, tracking and characteristics recording has been established based on imagery processing and time series analyzing. First, smooth sharp noise of the satellite image with mean filtering method and median filtering method. Then binary convert the preprocessed images, identify a MCS cloud regiment by marking and extracting the characteristic quantity, and get each target cloud regiment of time sequence. By computing the possible position, the target MCS, is checked if its characteristic matches with the stored information in area, strength, etc. Thus, the time sequence of the MCS cloud regiment is tracked automatically. The method is applied in MCS identifying, tracking automatically with characteristic statistics during the flood over Huaihe River in 2003, and the validation results show that this method has the ability of identifying MCS quickly each time, as well as tracking MCS of multi-time effectively.