Evacuation planning and shelter site selection are the most important function of disaster management for the purpose of helping at-risk persons to avoid or recover from the effect of a disaster. This study aims to propose a stochastic linear mixed-integer mathematical programming model for improving flood evacuation planning and shelter site selection under a hierarchical evacuation concept. The hierarchical evacuation concept is applied in this study that balances the preparedness and risk despite the uncertainties of flood events. This study considers the distribution of shelter sites and communities, evacuee's behavior, utilization of shelter and capacity restrictions of the shelter by minimizing total population-weighted travel distance. We conduct computational experiments to illustrate how the proposed methodical model works on a real case problem in which we proposed Thai flooding case study. Also, we perform a sensitivity analysis on the parameters of the mentioned mathematical model and discuss our finding. This study will be a great significance in helping policymakers consider the spatial aspect of the strategic placement of flood shelters and evacuation planning under uncertainties of flood scenarios.