Abstract: To address reduced renewable availability, increased emergency drainage demand, and constrained main-grid support under extreme rainfall, this paper proposes an active defensive dispatch method against meteorological risks for multi-energy microgrids. Based on warm-season meteorological data in Beijing from 2000 to 2023, the maximal information coefficient and Spearman's rank correlation coefficient are used to identify associations among compound meteorological factors. A total of 105 heavy rainfall events are reduced to 10 representative scenarios, from which parameterized operating boundaries for renewable availability, emergency drainage load, meteorologically sensitive load, and main-grid input are constructed. Aggregated regional insurance claim information from 2017 to 2023 is used to construct an actuarial spatial valuation factor, which is used solely as a post-dispatch regional valuation coefficient under fixed physical risk exposure. Under the same meteorological scenarios, baseline loss criteria, and physically feasible set, the risk-term scaling factors of conditional value-at-risk (CVaR) and GlueVaR are calibrated against the reference loss distribution of the expected-cost strategy, and three dispatch strategies based on expected cost, CVaR, and GlueVaR are formulated. Results show that, compared with expected-cost dispatch, the CVaR and GlueVaR strategies reduce expected load shedding by 18.25% and 18.70% and baseline GlueVaR by 8.69% and 8.91%, respectively, while increasing expected operating cost by 4.74% and 4.86%. Capacity-matching ablation experiments indicate that the operational role of the hydrogen energy chain depends not only on storage scale but also on multi-energy conversion pathways. The regional valuation factor changes only the monetary valuation of a given physical risk exposure and does not affect the physical dispatch results. The proposed method provides a reference for active defensive dispatch of multi-energy microgrids under extreme rainfall.