phases: A constructor function for class phase

In Issoufou-Liman/growingSeason: Estimating growing season metrics from seasonal data

Description

Takes vector (typically seasonal data such as rainfall time series) makes phases or seasons, and returns an object of class phases or seasons. phases, are here defined as a cycle of some sort that can be distinguished in the time series. The only important difference between phases and seasons is that seasons are phases having a peak.

Usage

phases(
  x,
  type = c("v_points", "peaks"),
  n_criticals = 1,
  steps = 2,
  ts_freq = 23,
  returned = c("ts_seasonal", "original")
)

seasons(x)

Arguments

x	a numeric vector which names are dates at which the data were acquired.
type	character string. either v_points or peaks depending on how x should be broken into phases. v_points (default) may make more sens for expression seasonal data.
n_criticals	Numeric. The number of points to be considered. Default to 1.
steps	Integer. The minimum number of points at either sides of each v-like point to be considered higher for it to qualify as v-like point. Default to 3
ts_freq	The frequence of time series (see frequency argument in ts). Default to 23.
returned	character string either ts_seasonal or original. if ts_seasonal (default), the returned phases data will be the seasonal component of the decomposed time series (see decompose). if 'original', the original data will be returned.

Value

an object of class phases comprise a list of:

• "phases" list of phases

• "criticals"index of the points separating the phases

Author(s)

Issoufou Liman

See Also

get_falls, get_phases.

Examples

## 2 years of a uni-modal pixel with complete without NAs
nam<- seq.Date(from = as.Date('2016-01-01'), to = as.Date ('2018-12-31'), by = 16)
dx11 <- c(1.30, 1.15,  1.50,  2.00,  2.01,  3.00, 3.20,  4.76,  3.50,  3.00,  2.40,  2.00,  1.50)
dx12 <-c(1.29, 1.1, 1.49, 1.99, 2, 3.1, 4.5, 4, 2.8, 2.5, 2.3, 1.6, 1.59)
dx1 <- c(dx11, dx12)
names(dx1) <- nam[1:length(dx1)]

## plotting the data
default_par <- par(no.readonly=TRUE)
layout(rbind(c(1, 1), c(2, 3)))
par(mar = c(2, 2, 1, 1))
plot(dx1, type = 'o', main = 'raw data') # note 2 phases = 2 years and the starting point.

## returning the phases as seasonal data.
y1 <- phases(dx1, ts_freq = 12)
y1
lapply (X = y1$phases, FUN = plot, type = 'o',
 main = 'phases extracted as a seasoanal component of ts object')

## returning the phases as original data
y2 <- phases(dx1, ts_freq = 12, returned = 'original')
y2
lapply (X = y2$phases, FUN = plot, type = 'o',
main = 'phases extracted as raw')
par(default_par)

## 2 years of uni-modal pixel with many Nas towards the end.
dx2 <- dx1
dx2[21:length(dx2)] <- NA
default_par <- par(no.readonly=TRUE)
layout(rbind(c(1, 1), c(2, 3)))
par(mar = c(2, 2, 1, 1))
plot(dx2, type = 'o')
## returning the phases as seasonal data.
y1 <- phases(dx2, ts_freq = 12)
y1
lapply (X = y1$phases, FUN = plot, type = 'o')

## returning the phases as original data
y2 <- phases(dx2, ts_freq = 12, returned = 'original')
y2
lapply (X = y2$phases, FUN = plot, type = 'o') # note 1 phase and the starting point
par(default_par)

Issoufou-Liman/growingSeason documentation built on Jan. 23, 2020, 3:43 a.m.