Abstract: To enhance microgrids' long-period off-grid operation capability under extreme disasters, a risk-aware optimization dispatching method is proposed for a hydrogen electricity coupling microgrid. Firstly, typhoon path simulation and distribution network fault probability assessment are conducted to perceive the islanding risk of microgrids in advance. Then, a multi-time scale optimization framework that couples long-period energy and day-ahead power dispatch is constructed. The energy optimization determines microgrids' long-period energy scheduling scheme, providing backup energy boundaries for day-ahead scheduling. The day-ahead optimization combines conditional value at risk and stochastic optimization methods to optimize hourly scheduling plans considering uncertainties. The proposed method can optimize the operation of microgrids over a longer period, leveraging the long-term energy storage of hydrogen and the flexible regulation ability of electrochemical energy storage. Simulation results show that the load loss of the proposed method is reduced by 98% compared with the traditional day-ahead scheduling method, significantly improving the ability of the microgrid to cope with extreme risks.