To analyze the regional characteristics of temperature changes in China, by using the daily mean temperature series from 1961 to 2002 of 435 weather stations in China, we set up the temperature fluctuant networks, calculate the topology statistics of every station, and present the special distribution graph of all stations. We find that the degrees of RRD, RDD, DRR, DDR, DrR are extraordinarily high, and the occurrence probability of temperature fluctuation represented by these 5 nodes is greater, which is helpful for forecasting the trend of temperature changes. Then,from the distribution graph of the degrees of these 5 nodes in all the 435 stations, we define them by the symbols as countrywide model, hypo-countrywide model, eastern model, blend model, and western with southern model. The coefficient between the degree distribution in every station and the national average degree distribution has a regional characteristics, which provide us with the feasibility for analyzing the regional characteristics in theory. From the regional characteristics of the national distribution of the average clustering coefficient, we define three kinds of significant regions of temperature changes in China: strong complex regions, weak complex regions and middle complex regions.