Province-level panel data for China (2000–2021) are interrogated to test the still-prevalent assumption that sustainable-development gains accrue monotonically to disaster-risk reduction. Non-parametric archetype analysis first decomposes the 16-goal system into nine distinct interaction topologies; non-linear regimes govern >60 % of all pairwise relations, with “unclassified weak” links acting as the dominant brake on aggregate CAGR(O/D). Embedding these evolving synergies within a refined National Risk Index reveals a significant U-shaped association between composite SDG performance and expected economic losses, inflecting at SDG Index ≈ 61. While community resilience scales linearly with development, social vulnerability rebounds beyond the threshold, driving the observed risk upturn. Nationally, SDG 1 (No Poverty) and SDG 4 (Quality Education) deliver the largest marginal risk attenuation; sub-nationally, the identity of leverage goals diverges—underscoring spatial heterogeneity. We therefore propose a dual-track governance architecture: macro-level surveillance of vulnerability tipping points coupled with locale-specific sequencing of root-cause leverage goals until resilience gains outpace exposure increments. The open-source archetype code and risk layers are transferable to other rapidly developing economies formulating integrated post-2030 and Sendai mid-term reviews.
 

Sustainable Development Goals,Disaster Risk Reduction,Non-linear Interactions,Multi-scale Analysis,Policy Implications