Background
Most SWMM-based LID evaluations rely on fixed AMC assumptions. In monsoon climates, soil moisture fluctuates dramatically between dry spells and prolonged wet sequences, so static AMC introduces systematic bias into LID effect evaluations.
Method
- Authentic GIS trunk sewer network (2,497 conduits) for Seocho/Gangnam-gu
- 2019–2020 hourly rainfall + dynamic Green-Ampt initial moisture deficit (IMD) coupling
- 4 LID types × 10–40% coverage × 15 scenarios
- XGBoost surrogate + SHAP attribution for variance decomposition
Key Results
| Metric | Value | |---|---| | Peak-flow shift from AMC alone | −5.6% to +6.2% | | Dynamic AMC vs AMC II baseline | −0.9% peak | | Improvement during 2020 extreme monsoon | up to +6.2% | | Annual runoff reduction at Mixed LID 40% | 39.6% | | Peak-flow attenuation at same scenario | 33.8% | | Unit efficiency at 10% coverage (max) | 98.3% |
Significance
That AMC variation alone produces peak-flow shifts equivalent to 10–15% LID coverage means static-AMC LID cost-effectiveness comparisons cannot distinguish intervention effect from model error. This framework provides quantitative, evidence-based guidance for LID planning in monsoon-dominated East-Asian megacities.