With the rapid increase in demand for the construction and development of island cities in the South China Sea, the urban heat island phenomenon in such cities should become a key factor to be considered in future urban planning. This paper took Sanya, China as a typical case, and long-term field experiments were conducted in the Mangrove Bay Area in summer and winter. An innovative urban green space cooling model was proposed, using the “green space cooling index” to quantitatively characterize the green space cooling effect, and aiming to minimize the intensity of urban heat islands. This paper studied the optimization method of green space planning and layout under the constraint of a centralized green space total area. Moreover, a genetic algorithm was adopted to optimize the calculation and the layout of the urban green space. The experimental results showed that the urban heat island intensity was more significant at night and was less effective in the daytime during summer. In winter, the urban heat island intensity had a greater effect in the daytime and was less influential at night. Finally, optimization results indicated that the average urban heat island intensity in summer was reduced by 8.8% under the optimal layout urban green space of 0.025 km2. The maximum reduction in heat island intensity occurred at 7:00 am (0.48 °C). When 0.0625 km2 urban green space was planned, the average urban heat island intensity index in summer was reduced to 0.27 °C, with a decrease of 20.5%.