[en] Today's increasingly interacting National Critical Infrastructures (NCIs) can tolerate most stochastic local disturbances, but they are very fragile under global disturbances, as the latter may either push the whole system into a critical state or reveal many unexpected hidden interdependencies; thus can induce or trigger cascading failures among all possible layers. This “robust yet fragile” duality is an inherent vulnerability of our modern infrastructures. Based on our research, as well as similar conclusions from other scientists, we find that weather related disasters will be more frequent under a changing climate. Since preventing all of these events will be costly and intractable, we propose an interdependency control strategy (ICS) that will maintain the survival of the most critical services, and compensate for this inherent vulnerability during emergency states. In this paper, we also propose a generalized adjacency matrix (GAM) to represent the physical interdependencies intra/inter various infrastructure networks. Based on computed results of GAM, number of islands in the network, influence domain(s) of each component, and the vulnerable section in the network can be identified. These features make ICS more effective and convincing. Moreover, a survivability index for isolated sub-networks is proposed and relevant measures for improving this index during the four phases of emergency management are described. (author)