README.md
In joaofgoncalves/phenexRaster: Apply phenex analysis to raster objects

phenexRaster

A set of ancillary functions for running phenex R package

Installation is easy you just have to run the following lines:


if(!("devtools" %in% installed.packages()[,1])){
  install.packages("devtools")
}

library(devtools)
install_github("joaofgoncalves/phenexRaster")

[[work in progress... ;-)]]

phenexRaster provides some ancillary tools to apply phenex functionalities to a \code{Raster*} object. To do this it applies in sequence the following functions: - modelNDVI: used to model, smooth and interpolate VI time-series), and, - phenoPhase used for Phenological Phase Extraction (calculation of 'phenometrics').

Typically, satellite time-series, such as those provided by MODIS, make available 8- or 16-day composites (e.g., MOD13Q1 16-day/250m data product) while phenex works with daily data. phenexRaster starts by constructing an annual data matrix with daily time-steps; then, values from each composite are injected in the matrix in positions corresponding to the reference Julian day of the composite. This makes it easier to smooth and interpolate values of vegetation indices using the modelNDVI function. Using this matrix we can then extract phenological phase indicators by applying the function phenoPhase.

These functions are experimental and errors may happen! In addition, the performance of phenoPhase may not be suitable to work with large datasets (npixels > 5E5 or so). A couple of limitations apply to this package:

Only complete years can be used (at least for now);
Only a single value can be passed to modelNDVI function: multipleSeasons, correction, and, method parameters; as well as for phenoPhase function: method and threshold parameters.