||Survivability of mobile networks has been a fundamental concern for network design and operation, especially in hazard-prone areas and countries such as Japan. Network survivability analysis is essential for effective disaster preparation and fast response. However, correlated network failures caused by the disaster propagation have not been well addressed in previous studies. Therefore, in this study, we investigate the network survivability by using a spatio-temporal data analysis during a large-scale natural disaster. Specifically, we consider the dynamics of disaster propagation by jointly analyzing essential geographical features from various sources, e.g., seismic intensity data, buildings and land usage data, base station location data, and many other data in related studies. We demonstrate our analysis with a synthetic earthquake scenario in Tokyo area, and validate our analysis by comparing to related studies. Our results could be useful in quantitatively evaluating the survivability of network design and various disaster countermeasures.