Severe Convection Weather Identification Algorithm And Its Implementation Based On Radar Data

The CINRAD has become one of the main equipments in monitoring the short-time severe convective weather, which can be used to identify the meso-scale and micro-scale weather systems, and then the weather warning is issued. It has an important significance for the meteorological disaster reduction.This paper firstly introduces the structure evolution of severe weather, the weather feature and explains the synoptic meteorology basis of the occurrence and development of the storm. And on that basis, this paper makes a contrastive study on the TITAN (Thunderstorm Identification, Tracking, Analysis and Nowcasting) algorithm and SCIT (Storm Cell Identification and Tracking) algorithm, which are used more frequently at present. Through the contrastive study, we can see that the TITAN algorithm has little computation workload and high recognition efficiency of storm outline; however, it doesn’t extract the internal parameters of storm and distinguish the subtle structure of storm cluster very well, which may result in the false merger. The SCIT algorithm can identify the strong storm center and isolate storm cell in a cluster of storms. But this method rejects the identification result of low threshold, which lead to the significant loss of the structure information of storm. This paper draws the advantages and disadvantages of these two algorithms, and then introduced the contiguous storm points for the identification of the two-dimensional storm component, controlling the identification of storm cell by the multi-threshold, and it can improve the false merger and split problem among the storm cells. Besides, the TREC algorithm and weighted extrapolation are used for tracking and forecasting respectively.In addition, the paper also discusses the quality control problem of radar data and describes the different echo by echo identification, horizontal structure of reflectivity, and horizontal texture of reflectivity etc. And according to the characteristic quantity of different echo by actual detection, a Logistic Regressive Differentiation Model is established to identify the abnormal propagation, air echo, precipitation echo, and non-precipitation echo, which can reduce the influence of radar data quality.The engineering design is based on the identification algorithm of strong convective storms and designed with Visual C++6.0advanced program language, which integrates the functions of processing, inverting and displaying raw data. Besides, it packages different product algorithms by DLL (Dynamic Linkable Library) and extend the features of an application. At the same time, it realizes the real-time browse of radar data at any azimuth in the geographical space by Google Earth API.At last, the present writer makes a discussion about the insufficient of storm identification and tracking algorithm, meanwhile, puts forward the next step of work plan.