A challenge for scalable time-varying volumetric data analysis and visualization is lacking the capability for the scientists to visualize, track and understand data evolution in a simple, controllable, and on-the-fly way. We introduce TransGraph, which is a graph-based representation of scalable time-varying volumetric data. By extracting transition relations, it provides a clear observation of time evolution, meanwhile, transGraph serves as a navigation tool to guide the scientists about the interesting volumetric regions, time steps and transition relations. In this approach, we first convert data to a set of states, each state is a spatial-temporal volumetric region. After deriving the transition relations among states, a graph is generated. Then we provide a set of interactions on the volume and graph, for example, brushing and linking, queries, and graph-based feature extractions so that the scientists could have a deep understanding of the original data. In this talk, I will explain how we convert a data set to states, and calculate transition relations. I will also talk about the approach to generate the graph. At last I will cover some interaction results and the work we have not finished.