China has recognized sponge city as a means to solve water shortage, combat inundation and water quality problems.The concept of sponge city is to implement low impact development (LID) at a large scale urban planning with multiple land use functions.It emphasizes the conservation and use of existing natural site features integrated with distributed, small-scale stormwater controls to closely mimic natural hydrologic patterns in residential, commercial, and industrial settings.Low impact development (LID) has gained place in urban planning due to its capability and beneficial effect in restoring the original hydrological cycle.The performance of LID alternatives can be varied substantially due to the climate condition.This study investigated the performances of a few LID alternatives on water balance and flooding control.A numerical model, USEPA SWMM5, was employed to run 10 years continuous rainfall events for this purpose.Two evaluation methods were proposed in this study including the efficiency index for water balance (IW) and performance radar chart (PRC).The investigation of the LID alternatives revealed that these LID alternatives function differently in for different land uses.For a large scale land development with multi-functional land uses, it therefore needs a tool to facilitate the LID implementations to optimize and enhance the restoration of hydrological cycle.A data mining approach was introduced in this study for this purpose and the methodology was applied to a new town development with an area of 45 km2.The case study demonstrates the efficiency and effectiveness of the methodology in optimizing the LID designs for a large scale land development.