Adaptive optimization of LID systems: spatial heterogeneity and rainfall intensity as drivers

ABSTRACT Graphical abstract in three parts. Top: two driving factors for LID configuration evolution — rainfall intensity (2‑, 5‑, 10‑, 20‑year) and spatial heterogeneity across five urban zones. Middle: BDP‑ES algorithm and SWMM‑LISFLOOD‑FP coupled model. Bottom: optimized LID layout evolving from binary to ternary system of PP, RG, and GR. To clarify how Low Impact Development systems should be optimally configured under urban spatial heterogeneity and varying rainfall intensity, this study proposes a deterministic optimization framework validated using the coupled SWMM-LISFLOOD-FP model. Taking Lianyungang City, China, as a case study, optimal LID layouts are found to adhere to a ‘location-specific matching’ principle and are evolved with rainfall intensity. Spatially, Commercial Districts prioritize rain gardens. Under low-intensity rainfall, Residential, Educational, and Public-Service Districts, as well as Green spaces, mainly use permeable pavements. As intensity increases (2- to 20-year return periods), Educational and Public Service Districts shift from permeable pavements to rain gardens, while the share of green roofs increases across all zones. The macro-system is evolved from a dual ‘infiltration-retention’ paradigm to a ternary system integrating ‘three-dimensional detention-terminal storage/regulation-source infiltration’. Post-optimization, hydrological performance improves significantly: runoff control rate remains >63.6%, peak runoff reduction >63%, junction overflow volume decreases by 79.3%–88.9%, and medium/deep inundation is substantially reduced. This enhanced resilience can be attributed to the synergy of ‘temporal hydrological attenuation’ and ‘spatial hydrological load redistribution’. The study provides a transferable, deterministic methodology for building stormwater resilience in heterogeneous urban spaces under climate stress.