Aiming at the performance degradation of multi-objective evolutionary algorithms caused by excessive parameters in optimizing spatial layout of low impact development(LID) facilities, the LID spatial layout optimization model of the sponge city construction area in Xiushan, Chongqing was established. Comparative analysis using multiple stepwise regression and SOBOL methods revealed the sensitivity of LID layout parameters to node overflow, identifying critical sensitivity parameters. The Pareto optimal front was subsequently refined based on sensitivity analysis outcomes to obtain the optimal LID spatial layout scheme. The results show that as the rainfall return period increases, the sensitivity of LID layout parameters in Xiushan’s sponge city area generally decreases due to saturation of runoff retention capacity. The sensitivity of LID layout parameters to node overflow is significantly influenced by drainage capacity of the drainage network in catchment areas, with parameters in regions with inadequate drainage capacity exhibiting remarkable sensitivity. Optimization of sensitive parameters substantially reduces evaluation times in NSGAⅡ algorithm while achieving a 90.29% reduction in node overflow under equivalent construction costs compared to preoptimization scenarios.