Enterprise AI Analysis

Smarter Spaces, Smoother Flows: Evaluating the Terrain Adaptability of LID Layouts via NSGA-II Optimization

The rapid expansion of impervious surfaces driven by urbanization has intensified the risk of urban waterlogging, posing a major challenge to sustainable urban development. Low Impact Development (LID) is widely recognized as an effective strategy to mitigate this issue; however, the adaptive mechanisms linking its spatial layout to complex terrain conditions remain insufficiently understood. Enabled by recent computational advances in the integration of hydrological simulation and AI-driven optimization, this study proposes a dual-path analytical framework that integrates forward simulation and AI-driven inverse optimization to examine the coupling relationship between LID layouts and terrain parameters. The framework first evaluates the hydrological responses of three LID layout types Dispersed, Clustered, and Centralized under various combinations of slope and imperviousness using forward simulation with the Storm Water Management Model (SWMM). It then employs the Non-dominated Sorting Genetic Algorithm II (NSGA-II), an evolutionary optimization method, to inversely identify the terrain parameter ranges within which each layout achieves optimal hydrological performance. The results indicate that: (1) both slope and imperviousness are key determinants of hydrological performance, exhibiting complex interactive effects; (2) a notable trade-off exists between performance potential and terrain adaptability. Specifically, the Dispersed layout achieves the highest performance but depends on gentle slopes, while the Clustered and Centralized layouts demonstrate greater adaptability across a wider range of terrain conditions. This study elucidates the coupling mechanisms between LID layouts and terrain characteristics, offering a quantitative, data-driven framework to guide the design of resilient and sustainable stormwater management strategies and spatial planning in cities with complex topography.