Effectively measuring cooling performance is important for evaluation of the thermal environmental regulation services of green roofs. Using the case study of the extensive green roof on the buildings of Jinling primary school, Nanjing, China, this study modeled the effects of green roof on local microclimate by using simulation-based ENVI-met software, and the simulation results were further validated based on the typical summer meteorological weather observations. Using meteorological data and ENVI-met simulation results, the cooling performance of the green roof was first analyzed from "point" and "surface" perspectives, similar to that by most previous studies, and then a novel analysis method was developed and applied to more comprehensively analyze the cooling performance from a 4D perspective. Accordingly, the hourly cooling effect and the spatio-temporal variations in the "cooling island" mass center were quantified, and simultaneously, the related factors potentially affecting the cooling performance of the roof were explained. The results showed that (1) the cooling effect of extensive green roof with Sedum lineare Thunb. was weaker during the daytime (6:00-18:00) than during nighttime (19:00-5:00 the following day). (2) The diurnal cooling effect was much more obvious when measured from a three-dimensional perspective than a two-dimensional one (ratio 3.1, 5.0, and 8.1 for the hourly cooling effects during daytime to that of nighttime from a point, surface, and three-dimensional perspective, respectively). (3) The mass center and intensity of the cooling island showed a pronounced spatial aggregation within the modeled area. The intensity of the cooling island was stronger during nighttime than during daytime. The mass centers gathered above the geometric center of the green roof during nighttime, whereas at nighttime, they scattered and shifted to the side of the bare roof. (4) The overall green roof-cooling performance indicated that the cooling effect was mainly affected by the properties of the underlying surface, the physiological characteristics of the vegetation, and the local micro-meteorological factors. Compared with that of the traditional method, the measurement from the four-dimensional perspective, achieved a more objective assessment of the green roof-cooling effect. The proposed novel method will be extremely helpful in the cost-benefit analysis and of green roof-construction projects.