R: Wavelet Sample Spectrum

wsp {sowas}

R Documentation

Wavelet Sample Spectrum

Description

This funtion estimates the wavelet spectrum of a time series object with the Morlet wavelet.

Usage

wsp(ts, s0 = 1, noctave = 5, nvoice = 10, w0 = 2*pi, sw = 0, tw = 0,
swabs = 0, siglevel = 0.95, critval = NULL, nreal = 1000, arealsiglevel
= 0.9, kernel = 0, markt = -999, marks = -999, logscale = FALSE, plot =
TRUE, units = "", device = "screen", file = "wsp", color = TRUE, pwidth
= 10, pheight = 7, labsc = 1, labtext = "", sigplot = 3)

Arguments

`ts`	time series object to be transformed
`s0`	lowest calculated scale in units of the time series
`noctave`	number of octaves
`nvoice`	number of voices per octave
`w0`	time/frequency resolution omega0
`sw`	length of smoothing window in scale direction is 2swnvoice+1
`tw`	length of smoothing window in time direction is 2stw+1
`swabs`	length of smoothing window in scale direction at scale s is 2*swabs+1
`siglevel`	vector of significance levels for pointwise test, e.g. c(0.9,0.95), default 0.95.
`critval`	matrix of critical values calculated e.g. by critical.valuesWSP; each row contains critical values for the corresponding chosen significance level.
`nreal`	number of realizations to estimate critical values for the corresponding significance values, default 1000
`arealsiglevel`	significance level of the areawise test; currently only for siglevel=0.9 and for arealsiglevel=0.9 possible, i.e. 90 percent of the area of false positive patches is sorted out
`kernel`	bitmap of the reproducing kernel; if not provided, it will be calculated during the areawise test
`markt`	vector of times to be marked by vertical dotted lines; when set to -999 (default), no lines are plotted.
`marks`	vector of scales to be marked by horizontal dotted lines; when set to -999 (default), no lines are plotted.
`logscale`	when TRUE, the contours are plotted in logarithmic scale
`plot`	TRUE when graphical output desired
`units`	character string giving units of the data sets. Default: ""
`device`	"screen" or "ps"
`file`	character string giving filename of graphical output without extension
`color`	TRUE (default): color plot, FALSE: gray scale
`pwidth`	width of plot in cm
`pheight`	height of plot in cm
`labsc`	scale of labels, default: 1, for two-column manuscripts: 1.5, for presentations: >2
`labtext`	puts a label in upper left corner of the plot
`sigplot`	0: no significance test plotted, 1: results from pointwise test, 2: results from areawise test, 3: results from both tests

Details

A pointwise significance test is performed either by providing critical values (e.g. from critical.valuesWSP) or by Monte Carlo simulations. If critval and either nreal or siglevel is zero, no significance test is performed. If arealsiglevel=0.9, an areawise significance test is performed, that sorts out 90 percent of the area of false positive patches. The cone of influence is marked by black lines. Values outside the cone of influence should be interpreted very carefully, as they result from a significant contribution of zero padding at the beginning and the end of the time series. For a better visualization, additional dotted lines marking distinct times or scales might be plotted by providing the vectors markt and marks. In the linear plot, values are normalized such that the highest spetral power equals one. In the logarithmic plot, values are normalized, that the lowest value equals 1. The function returns an object of type "wt", that might be directly plotted by the plot function.

Value

`modulus`	here: wavelet sample spectrum of dimension [length(ts)]x[nvoice*noctave+1]
`phase`	not used
`s0`	lowest calculated scale in units of the time series
`noctave`	number of octaves
`nvoice`	number of voices per octave
`w0`	time/frequency resolution omega0
`time`	vector of times of length(intersection of ts1 and ts2)
`scales`	vector of scales of length nvoice*noctave+1
`critval`	matrix of critical values
`at`	time/scale matrix of areawise significant patches
`kernel`	bitmap of reproducing kernel

Note

Author(s)

D. Maraun

References

D. Maraun and J. Kurths, Nonlin. Proc. Geophys. 11: 505-514, 2004

Examples

##
data(nino3)

wspnino3 <- wsp(nino3,s0=0.5,noctave=5,nvoice=10,nreal=100,units="years",arealsiglevel=0)

[Package sowas version 0.93 Index]