The incidence of extreme weather is increasing yearly, threatening the safe and stable operation of power grids. To solve the problems of limitations that exist in extreme weather historical data sampling modeling, inaccurate extreme weather prediction, and insufficient absorption capacity of the power grid to cope with natural disaster defense, this paper proposes a collaborative emergency defense strategy for offshore wind farm clusters that takes into account the spatial-temporal characteristics of typhoons. First, based on the Batts gradient wind model, it analyzes the spatial-temporal characteristics of a typhoon and the influence mechanisms, and quantifies the coupling relationship between the physical information on the typhoon and the output of offshore wind farms. Then, a two-stage stochastic optimization model for offshore wind clusters with multiple time scales considering the spatial-temporal characteristics of typhoons is proposed with the objective function of minimizing the emergency defense cost within the deviation assessment interval. The first stage of the model is based on the typhoon forecast information to customize the defense resource pre-dispatch strategy including conventional unit standby output, pumped storage unit start/stop plan, energy storage plant output, and planned wind curtailment/load shedding. The second stage incorporates the typhoon forecast uncertainty into the decision-making process to formulate the emergency defense adjustment strategy based on the flexible distributed resources. Finally, a case study verifies the effectiveness and economy of the proposed emergency defense strategy.