detect_forages: Detect forages
In ranghetti/sen2rts: Build and Analyse Sentinel-2 Time Series

detect_forages

R Documentation

Detect forages

Description

Experimental function to detect forages (time series characterised by a high "instability"). Already detected cycles are labelled as "forage" (adjacent ones are dissolved) basing on arguments diff_thr and ncuts_thr (see).

Usage

detect_forages(
  ts,
  pheno,
  win = NULL,
  diff_thr = 10,
  ncuts_thr = 10,
  sliding = 90,
  reldiff = TRUE,
  byyear = TRUE
)

Arguments

`ts`	Time series in `s2ts` format (generated using `fill_s2ts()`).
`pheno`	Cycle allocation (data table generated by `extract_pheno()` or `cut_cycles()`).
`win`	List with the allowed ranges for the dates of start of cycle (see `pop_seasons` for details about the format).
`diff_thr`	First threshold to detect forages (TODO improve description)
`ncuts_thr`	Second threshold to detect forages (TODO improve description)
`sliding`	Length (in days) of the sliding window in which `diff_thr` and `ncuts_thr` must be evaluated.
`reldiff`	Logical: should argument `diff_thr` refer to relative (default) or absolute values?
`byyear`	Logical: compute aggregation by year (default) or globally (this should make sense in case of permanent crops).

Value

A data table in the format of the input pheno, in which adjacent forage cycles are merged. An additional logical field forage discriminates forages from non-forages.