A Prediction Scheme For The Frequency Of Summer Tropical Cyclone Landfalling Over China Based On Data Mining Methods

The tropical cyclone (TC) path largely determines the effect of TC and TC regional influence scope. The frequency of TC directly affects the development of national coastal economic zone economy construction, society and people’s life and property security. TC activity is a very complicated natural phenomenon, so for activities of the TC research has always been meteorology has been one of the hot issues. Mining Based on TC track data set, not only can find the relationship between the frequency of TC and known climate signal, is more important for the production of human life, the construction of the state and society to provide more favorable security and support. The data mining technology in history landing path of China’s TC classification, and many people known climate signals using classification and prediction algorithm in two kinds of decision tree algorithm on each path of the frequency of TC for the classification and prediction and scientific, simple, intuitive classification rule set based on, the better to reveal the relationship between all kinds of path frequency of TC and people known climate signal, for predicting the frequency of TC provides a new idea.Specifically, this paper explore and study from the following three aspects:1. This paper deeply understanding the focus and difficulty in the present research in the field of tropical cyclone, from the data mining point of view, the design of the landing China TC path classification model and all kinds of path frequency of TC classification and prediction model. Through to data mining clustering algorithms and classification and prediction algorithm knowledge of in-depth study, we have chosen the appropriate algorithms for classification of TC tracks the FMM algorithm for model establishment; and for frequency of TC classification and prediction of the two kinds of decision tree algorithm:cart and C4.5 algorithm, using the two algorithms of path frequency of TC model construction and validation. Play the meteorology, data mining, computer science and interdisciplinary advantages.2 this paper uses clustering algorithm to the TC landing path Chinese were divided relatively objective. Through the K-means algorithm and FMM algorithm two categories are often used by scholars to divide the TC path of the clustering algorithm of learning research and comparison, this paper argues that the FMM algorithm for the TC path of the classification of more scientific,. Based on empirical observation method, history landing China TC path can be divided into three types, and has carried on the analysis to each kind of path summer autumn and summer frequency and life history characteristics, found each landing China TC life history distribution have different characteristics. But the frequency of various types of TC is complex, and it is difficult to find the law.3. In this paper, try using the cart algorithm and C4.5 algorithm through a variety of summer climate signal data on all summer time path frequency of TC is too much separately carried on the classification and prediction, achieved better classification and prediction results, and generates a simple, scientific and effective rule set, convenient intuitive access rules between the TC frequency and the climatic signal.The above work not only enriches the research of tropical cyclone, but also provides a new idea for the combination of data mining and meteorological research problems.