Portal de Periódicos da CAPES

Phenopix: A R package for image-based vegetation phenology

2016; Elsevier BV; Volume: 220; Linguagem: Inglês

10.1016/j.agrformet.2016.01.006

1873-2240

Gianluca Filippa, Edoardo Cremonese, Mirco Migliavacca, Marta Galvagno, Matthias Forkel, Lisa Wingate, Enrico Tomelleri, Umberto Morra di Cella, Andrew D. Richardson,

Plant Water Relations and Carbon Dynamics

In this paper we extensively describe new software available as a R package that allows for the extraction of phenological information from time-lapse digital photography of vegetation cover. The phenopix R package includes all steps in data processing. It enables the user to: draw a region of interest (ROI) on an image; extract red green and blue digital numbers (DN) from a seasonal series of images; depict greenness index trajectories; fit a curve to the seasonal trajectories; extract relevant phenological thresholds (phenophases); extract phenophase uncertainties. The software capabilities are illustrated by analyzing one year of data from a selection of seven sites belonging to the PhenoCam network (http://phenocam.sr.unh.edu/), including an unmanaged subalpine grassland, a tropical grassland, a deciduous needle-leaf forest, three deciduous broad-leaf temperate forests and an evergreen needle-leaf forest. One of the novelties introduced by the package is the spatially explicit, pixel-based analysis, which potentially allows to extract within-ecosystem or within-individual variability of phenology. We examine the relationship between phenophases extracted by the traditional ROI-averaged and the novel pixel-based approaches, and further illustrate potential applications of pixel-based image analysis available in the phenopix R package.