Using the daily main observational temperature data of 435 stations in China from 1961 to 2002, temperature change on 9-day scale has been compiled in network. Research on the connection between nodes and items in bipartite graph temperature (BGT) network, reveals the temperature change on 9-day scale and the topological statistics in the space. The nodes of RRRD, RrDD, eeed, DRRD and DDRR have remarkably high degree, which is helpful to predict the temperature change on 9-day scale. Calculation of the topological parameters of this network, including degree distribution and clustering coefficients, shows the normal school character of the bipartite graph model temperature network. The distribution of nodes’ degree diversity in each item presents quadruple type character, and has similar characteristics as the complex regions defined by the fluctuant temperature network(FT network), displaying the background information of temperature change in the region by this two kind of network respectively. Thus, temperature network modeled by bipartite graph model present a possible and available approach to research the characteristic and rule about temperature change from the combination of the time and space scale.