MtsPlot: Multiple time series plot
In greenbrown: Land Surface Phenology and Trend Analysis

Description Usage Arguments Details Value Author(s) See Also Examples

Allows to plot multiple time series with different y-axes into one plot. This function is an alternative approach to plot trends than plot.Trend. It is especially suited to plot trends of different variables with different units in one figure.

MtsPlot(x, xlim = NULL, 
    rge = c(0, 1), ylim = NULL, 
    col = "blue", lwd = 1, 
    add = FALSE, axis = TRUE, 
    ylab = "x", axis.pos = "l", 
    axis.col = col, axis.lim = NULL, 
    trend = TrendAAT, 
    trend.period = NULL, 
    trend.text = 1, unit = "", 
    text = "", unc = "polygon", 
    ensfun = mean, ...)

`x`	time series that should be plotted. If `x` has multiple columns, uncertainties can be plotted as individual 'lines' or as 'polygon' (see 'unc').
`xlim`	range of x-axis
`rge`	range of y-axis in which the time series should be plotted.
`ylim`	`axis` limits of the y-axis for the data in `x`
`col`	color of the time series and corresponding `axis`
`lwd`	line width of the time series
`add`	initialize a new plot if FALSE or adds time series to existing plot if TRUE
`axis`	draws a y-axis if TRUE
`ylab`	label for y-axis
`axis.pos`	position of y-axis: l (left) or r (right)
`axis.col`	color for the `axis`
`axis.lim`	limits (min/max) for the `axis`
`trend`	compute and adds a `trend` line using `TrendText`. Set to NULL for no trends.
`trend.period`	define a sub-period (e.g. c(1990, 2015)) to compute trends for the full time series and this period
`trend.text`	type of `trend` `text` to plot: 1 (`trend` in %), 2 (`trend` slope with `unit`), 3 (`trend` slope without `unit`), 4 (p-value from Mann-Kendall `trend` test), 5 (only if `x` has multiple columns: p-value Wilcoxon rank sum test). In case of other values, no `text` will be plotted. In case of trend.text=5, the Wilcoxon rank sum test is used to test if the ensmble of `trend` slopes for each column in `x` is significant different from 0.
`unit`	`unit` of the values in `x` (used for `trend` `text`)
`text`	additional `text` that should be plotted next to each time series
`unc`	if `x` has multiple columns, plot uncertainty ranges as 'polygon' or line'?
`ensfun`	if `x` has multiple columns, how to compute the main estimate of the emsemble?
`...`	further arguments to `plot.default`

This function plots a scatterplot.

A list with information about the estimated trend.

Matthias Forkel <matthias.forkel@tu-dresden.de> [aut, cre]

TrendText, plot.Trend

ndvi <- aggregate(ndvi, FUN=mean)

# default plot
MtsPlot(ndvi, ylab="NDVI")
MtsPlot(ndvi, ylab="NDVI", text="GIMMS")
MtsPlot(ndvi, rge=c(0.4, 1)) # plot time series in (0.4-1) range of y-axis
MtsPlot(ndvi, rge=c(0.5, 1), ylim=c(0,1)) # use (0-1) range for y-values and 
  # plot in (0.5-1) range of y-axis

# plot with uncertainty ranges
ndvi2 <- cbind(ndvi, ndvi * 1.3, ndvi * 0.7)
MtsPlot(ndvi2, rge=c(0.5, 1)) # uncertainty as polygon
MtsPlot(ndvi2, add=TRUE, rge=c(0, 0.5), unc="line", col="red",
        axis.pos="r") # uncertainty as line

# # plot multiple time series
# ndvi2 <- ndvi * 2 # scaled NDVI
# ndvi3 <- ndvi^2 # squared NDVI
# par(mar=c(3, 3, 2, 3))
# MtsPlot(ndvi, rge=c(0.66, 1), ylab="NDVI")
# MtsPlot(ndvi2, add=TRUE, rge=c(0.33, 0.66), axis.pos="r", ylab="NDVI*2", col="red")
# MtsPlot(ndvi3, add=TRUE, rge=c(0, 0.33), axis.pos="l", ylab="NDVI^2", col="purple")
# 
# # options for trend
# MtsPlot(ndvi, ylab="NDVI", trend=NULL) # no trend
# MtsPlot(ndvi, ylab="NDVI", trend.period=c(1982, 1996)) # compute trend in subperiod
# MtsPlot(ndvi, ylab="NDVI", trend.period=c(1982, 1996), 
#         unit="NDVI", trend.text=1) # text: trend in % (default)
# MtsPlot(ndvi, ylab="NDVI", trend.period=c(1982, 1996), 
#         unit="NDVI", trend.text=2) # text: trend in units
# MtsPlot(ndvi, ylab="NDVI", trend.period=c(1982, 1996), 
#         unit="NDVI", trend.text=3) # text: trend without unit
# MtsPlot(ndvi, ylab="NDVI", trend.period=c(1982, 1996), 
#         unit="NDVI", trend.text=4) # text: p-value only
# 
# ndvi2 <- cbind(ndvi, ndvi * 1.3, ndvi * 0.7)
# MtsPlot(ndvi2, ylab="NDVI", trend.period=c(1982, 1996), 
#         unit="NDVI", trend.text=5) # text: p-value only