As mentioned before, many thermal processes are described by DPSs, working in a large-scale operation region. In each region, it has special nonlinear dynamics due to specific relative position with heat sources. Achieving a global dynamic model of this kind of processes is extremely difficult due to different local dynamic features. Here, a graphic relation-based spatiotemporal fuzzy model is proposed to reconstruct the large-region DPSs. First, a spectral clustering strategy is developed to divide the large-scale spatiotemporal region into several local regions. For each local region, the SBFs are extracted to present the energy exchange on space. To reflect the global spatial feature, an incremental fusion approach is designed using fuzzy algorithm integrating these SBFs to form a global SBF. Then, temporal dynamics is obtained by projecting the spatiotemporal data on this global SBF and characterized by a fuzzy model. On basis of the global SBF and temporal model, the spatiotemporal model is constructed for the process with large-scale operation region. Finally, model effect of this method is demonstrated by theoretical analysis and experiment.