Sudden leakage has become a growing threat to public safety in cities. In this paper, using the emergency response to sudden leak accidents, which is a practical problem threatening public safety in cities as the entry point, we studied the simulation technology of sudden gas diffusion process on city-block-scale. Focused on the leakages that threaten urban public security, we suggested a point source gas diffusion model suit for city-block-scale and constructed a 3D spatio-temporal gas diffusion simulate model coupled with GIS. We use it to simulate the extent of the diffusion area for pollution monitoring and data analysis, and virtually reason the accident's damage degree to public and the subsequent development. So that the accident response support decision system can be constructed, the evacuation path and time can be calculated, the emergency rescue plan can be optimized. The most remarkable characteristic of the emergency rescue problem is the time urgency that decision-makers should propose emergency rescue plan within the shortest possible time, and that adequate rescue vehicles, personnel, and supplies would be mobilized and arrive at the accident sites at the fastest speed to carry out rescues. Faced with massive sudden leakages, it is very essential to optimally distribute the emergency rescue points for being able to effectively control disaster and to successfully carry out rescue activities. According to the spatial distribution of the emergency rescue points around the accident areas, we explored the algorithm for seeking the best path of aid resource allocation, obtaining the optimal solution to the distribution of the emergency rescue points through dynamic optimization strategies.