Retrieval of cherry flowering phenology on Flickr and YouTube: a case study along the Tarumi railway, Gifu, Japan

Abstract

To further develop the accuracy of monitoring cherry flowering dates, we require phenological records from multiple points in multiple years at the catchment scale, as well as conventional in situ phenological observations, phenological data published on the Internet, and analysis using statistics of Internet search engines. We tried to detect the dates of cherry flowering phenology along the Tarumi railway, Gifu Prefecture, Japan, by using Flickr, an image hosting service, and YouTube, an online video sharing and social media platform. We detected full bloom of Cerasus ×yedoensis and Cerasus jamasakura mainly at cherry blossom viewing spots (around some train stations) on Flickr and at both viewing spots and multiple points in the landscape along the railway on YouTube. Despite local climatological differences along the railway, the detected full blooming dates corresponded not only with each other, but also with the full bloom period of Neodani Usuzumi-zakura (Cerasus itosakura), a famous tree with long-term detailed records. We could not detect the date and location in many photographs on Flickr or in any videos on YouTube. However, the usefulness of both platforms is supported by the facts that we can evaluate the year-to-year variability of full bloom dates, especially at cherry blossom viewing spots, and detect flowering phenology even in a non-photogenic landscape. By applying our method to other railways, we expect to collect multi-year records of plant phenology dates at multiple points that cannot be collected by older methods.