Current studies on the evolution mechanisms of major natural disasters have largely been conducted on the basis of a single type of typical disaster event, while research from a broader spatiotemporal perspective remains limited. To address this gap, complex network theory was adopted in this study as the primary analytical framework. A three-layer complex network-based representation model for major natural disasters was constructed, and its effectiveness was validated using the Louvain algorithm, thereby enabling a comprehensive characterization of disaster risk factors across large spatial and temporal scales. On this basis, a link prediction method grounded in the proposed disaster network representation model was further developed to forecast the evolving risk situation of major natural disasters. Comparative analysis with the traditional Logistic prediction approach showed that the proposed method exhibits clear superiority in predictive performance. The main contribution of this study lies in the integration of complex networks and link prediction to develop both a network representation model and a risk situation forecasting method for major natural disasters in order to study the evolution of major natural disaster risks on a large spatiotemporal scale，which provides a new theoretical foundation and methodological support for disaster risk prevention and control.