Search

• Detailed social network site image survey of spatiotemporal ecological phenomena. • The seasonal bloom of Japanese cherries is revealed at a national scale. • A subtle autumn cherry bloom is detected using a decade of data. • The full bloom date of cherry blossoms is estimated from social network site images. Understanding how changing climatic conditions are impacting flowering plants typically requires intensive effort and expense to sample a local site regularly over long periods of time. The logistics of organising detailed surveys of an extensive area to provide a wider perspective are even more inhibitive. Data on flower bloom patterns across areas stretching hundreds or thousands of kilometres, and with a temporal resolution down to 1 or 2 weeks, is nevertheless very valuable, should it be feasible to collect. To understand the potential for contemporary data to record such flowering patterns, we studied Japan, a country where cherry (Sakura, 桜) flower viewing (Hanami) is a national cultural practice stretching back hundreds of years, and in which contemporary citizens and visitors commonly photograph blossoms to share on social network sites (SNS). We employed the big data this activity creates, within an iEcology framework, by collecting images from the SNS Flickr over the decade 2008–2018. We developed a custom filtering pipeline to validate this extracted data against established databases of historical flowering times. Our results reveal unprecedented detail of the spatiotemporal pattern over which cherry blossoms seasonally sweep from southern to northern Japan during a 12 week period. They also were sufficiently sensitive to reveal a subtle out of peak season bloom. This novel approach and data source therefore provides a simultaneously broad and detailed perspective that communicates the seasonal ecological phenomenon of cherry tree flowering. [ABSTRACT FROM AUTHOR]