Precise irrigation of agro forestry is imminent with the shortage of water resources in the world and the development of modern agro forestry and information science. One of the effective means to realize precise irrigation is to improve the control method and control accuracy. In order to study the effective control strategy for the irrigation of urban green land, an irrigation control model with optimal control theory was built. To testify the model, a dual-layer sensor was used to obtain the soil moisture in depth of two layers (20cm and 40cm) simultaneously. The experiment was conducted in an urban green land of Beijing. With the same climatic environment, the traditional irrigation control model with Bang-Bang control method and the optimal irrigation control model were compared. After comparative study in two months of 2011, the experiment result indicated that, comparing with Bang-Bang control, the optimal control saved water remarkably (approx. 25%) in the same weather ,soil and grass management conditions and ensured the lawn grass meeting the landscape requirements.