Small tropical lagoons are highly susceptible to eutrophication due to restricted hydrodynamic exchange and intensive land-based inputs. However, translating theoretical “environmental capacity” into actionable, spatially explicit control schemes remains a critical challenge. Taking the Xiaohai Lagoon in Hainan as a case study, this research establishes an integrated assessment framework coupling a two-dimensional hydrodynamic-water-quality model with grid-scale capacity accounting and multi-scenario load regulation. By integrating multi-source data, we quantitatively characterized the spatial heterogeneity of COD, DIN, and DIP and their remaining capacities. The results indicate: (1) Water quality exhibits a significant “inlet surplus-inner basin deficit” dual gradient. Under Class Ⅲ water quality targets, DIN is at critical saturation, while DIP acts as an “absolute rigid constraint” with zero remaining capacity; under Class II water quality targets, widespread nitrogen and phosphorus overload leads to capacity exhaustion. (2) Scenario simulations reveal that while hydrodynamic optimization engineering can release physical capacity, it has a “ceiling effect,” and the benefits are easily “offset” by incremental loads from socio-economic growth. (3) Only under a “deep source reduction” scenario coordinating the watershed and lagoon can the system breach physical thresholds, achieving Class II compliance in approximately 87% of the water area. The proposed “Accounting-Zoning-Reduction” framework facilitates a paradigm shift from gross total load control to precision governance characterized by “source reduction, spatial matching, and process optimization”, offering a transferable technical pathway for similar semi-enclosed coastal waters.