In coastal cities, population and property are concentrated in small areas, with abundant resources and convenient transportation, also with potential tsunami risk, such as the tsunami disasters of 2004, 2010 in Indonesia. Coastal area citizens need to evacuate to a safe place as soon as tsunamis occur. The prime evacuation time is very critical for them, but it is delayed in practice by complex information transfer processes. In recent years, spatial information has become an important resource used in dynamic decision support for emergencies, and smart phones have become a primary social communication device during interactions in emergencies. This paper outlines the design and development of a prototype Geographical Information and Mobile Social Network Dynamic Decision Support System (GIMSN-DDSS) that integrates geographical information software with mobile phone technology and the social communication network, Twitter. The actors include government policy makers, policy managers or highly influential social leaders in local communities, and policy executors or urban citizens influenced by disaster as Twitter followers. The main system functions include dynamic disaster risk analysis and diffusion of results, real-time detection of environmental risk, real-time analysis of possible evacuation routes and transmission of evacuation strategies to community residents. This system is designed as a field experiment in Padang, Indonesia, to help public officials design tsunami risk maps with timely evacuation routes and transmit these maps to influential leaders in small neighborhoods that are exposed to tsunami risk. Each neighborhood leader would then tweet the detailed route to citizens that follow his Twitter hashtag. This communications mechanism could offer evacuation guidance at the community scale of social networks directly from the early warning system.