Urban waterfront green space has become a space carrier of frequent physical activities because of its highly open and unique water resources. In digital media, network resources bring new cognition to the research of waterfront space and activities. This paper takes the waterfront green space usage pattern for all seasons in Harbin as the research object. Based on Sina Weibo, the data onto visitors and the types and intensities of physical activities are extracted，and applied to the GIS map to explore the spatial and temporal distribution and summarize the influence factors and spatial composition. The results show that: physical activity significantly tends to differ from weekends, holidays, and certain seasons; individual and group activities have different spatial preferences; climate and precipitation conditions affect the time selection of activities; the hydrophilic degree of the site has a significant impact on the type of activities, and the spatial form can effectively guide the social scale of group activities. In summary, according to the frequency, intensity, and scale of activities, this article proposes five blue space composition models based on physical activity, such as landscape comfort, walking through, entertainment experience, event triggering, and balanced development. It also provides a new way of accurately promoting the potential health of urban blue space.