Detection of phenological features of an evergreen broadleaved tree by analysis of fractal dimension

The development of phenological observations of evergreen broadleaved trees is required to deepen our understanding of ecosystem functions and services and of biodiversity under climate and societal changes. Toward that end, we examined the relationship between daily fractal dimension D values calculated from long-term timelapse canopy-surface images of Castanopsis sieboldii and the appearance and disappearance of plant organs and changes in their structure over time. The daily D values based on the size of outlines of plant organs (such as leaf buds, flowers, and new leaves) changed seasonally in accordance with developments such as leaf flush and flowering. The seasonal patterns of change differed between a year with many flowers and years with few to no flowers. Despite the need for further long-term daily timelapse images of various evergreen broadleaved trees to allow the extraction of outlines of smaller organs, our findings suggest that the analysis of daily D values is useful for detecting details of flowering and of leaf flush and senescence phenology, focusing on temporal changes in the structures of each organ.