Construct Meteorological Observation Network And Research On Key Features

Complex networks are widely applied to various fields, its related theory also provides help for interdisciplinary research on its complexity, such as the; Internet, biological networks, social networks, etc. At the same time, with the growing of new information release and access, the data are grew and accumulated at an unprecedented rate. How to make use of knowledge discovery based on big data becomes a research focus. In the field of the weather, because the weather system has the characteristics such as complexity and chaos, the planning has great randomness and meteorological stations. It is difficult to assess the relationship between the sites. So we proposed a meteorological observation network properties research method based on complex network.In this paper, we first preprocess the meteorological data, on the basis of guarantee the quality of data, we build directed and undirected networks respectively. And then we divide the nodes in the network and application of the control of complex network to the meteorological network. Finally, we put forward a method to evaluate the importance of nodes and improved the computing method of network efficiency and vulnerability.The main work of this paper has been summarized as follows:1. The construction of meteorological observation network.We constructed undirected network and directed network respectively in this paper. We constructed the undirected network with linear correlation coefficient method and build the directed network based on symbol sequence.2. Research the key features on meteorological observation network.In this paper, the research on the key features includes two parts. First, due to the fault of measurement equipment and other factors, measurement error can affect the clustering results. If we take the error factors into consideration, we can produce much better results, so we proposed a clustering algorithm with error based K-means. And then we make a new definition to evaluate the importance of nodes based on the controllability research and node partition. Second, According to the characteristics of various meteorological networks, in order to adapt to the evaluation of meteorological network, we redefined the efficiency and vulnerability assessment method.