In response to the serious methodological dichotomy of early seismic damage assessment and subsequent post-earthquake optimisation of logistics networks; To establish an integrated Spatial-Temporal earthquake-risk Assessment and Resource Allocation Paradigm. Traditional analysis paradigms of disaster operation research always separate hazard evaluation from the allocation decision; thus, deployed countermeasures often lack sufficient robustness against deep uncertainty due to information lag and potential system cascade failure risk. To address this epistemic rift systemically, the proposed mathematical architecture integrates a spatial-temporal stochastic ground-motion-field module that can describe anisotropy in spatial correlations and aftershock-evolution dynamics over time; It also links together a highly dynamical cascade of damage propagation model. Map inter-related functions’ decline within the urban system to build a probabilistic Directed Acyclic Graph (DAG) that continuously assesses compounding paths of infrastructure failures. The repeated vulnerability assessment is conducted by successively applying Bayes’ theorem to merge actual situation observations continually; therefore, the estimated amount of losses needs modification precisely when taking urgent measures. Based on the continuously adjusted dynamic risk environment, a multi-objective stochastic resource-scheduling optimiser is deployed here. Within the two-stage stochastic programming framework that is mathematically rigorous, it optimises simultaneously to minimise expected human casualties, mitigate cascadeeconomic depreciation; Accelerate resource deployment latency in response to strict enforcement of chance constraints to ensure guaranteed service level. Across four different metropolis-scale seismic datasets, computational implementations have shown significantly better performance than existing baseline methods. It is empirical evidence that can lower the probability of fatal injuries by 23.0% and increase resource use efficiency in production by another 15.0 per cent. Furthermore, the timely transmission of information can achieve an additional tenfold improvement (10×) increase in total casualties reduced compared to the advanced deterministic and sequentially integrated models system; Has been scientifically verified by a complete set of non-parametric ranked-sum test. Therefore, an integrated approach is proposed to build a rigorous quantification theory base and dynamic adaptability foundation for city-wide disaster mitigation exercises; The basis can be used as scientific guidelines in practice.
