In order to analyze past and future loss caused by natural disaster, it is important to know potential loss which could be brought about if the spatial and temporal scenario of disaster occurence were changed. Spatial and temporal distribution of human activity is one of the important factors which can influence on the potential loss. Especially this research focuses on the fact that the human activity is related to an SAR activity after disaster, and we developed the SAR model considering people's behaviors after the earthquake, characterized by age, residential location, and spatial and temporal activity location. Then, an SAR simulator system is constructed to simulate potential losses of human lives depending on both the time of the occurrence of the hazard and the place of human activities by combining the developed SAR model and spatial-temporal database. As a result of case study in Nagata, Chuo in Kobe, influence of spatial temporal distribution of human activity was estimated quantitatively.