multitaperTrend function

Estimate Linear Trend using Multitaper Techniques

Estimate Linear Trend using Multitaper Techniques

Estimate linear trend using inverse spectrum estimation, with the spectrum being computed via multitaper. This technique has improved spectral properties when compared to the least-squares approach. Returned values from this function include the intercept, slope, and centered time array.

multitaperTrend(xd, B, deltat, t.in)

Arguments

  • xd: Contiguous time series to be detrended.
  • B: Bandwidth to use in estimating trend in physical units; corresponds to NW via equation NW=BT, where N and W are the usual Slepian definitions, and T is the total time elapsed, i.e. T = N*deltat.
  • deltat: Time step for series xd, also used in computing T.
  • t.in: Time array, used in accurately estimating the slope.

Examples

x <- 1:101
y <- 1.0 + 0.5*(x) + rnorm(n=101,mean=0,sd=2)
vars <- multitaperTrend(xd=y, B=0.05, deltat=1.0, t.in=x)
plot(x,y,type="l")
lines(x,vars[[1]]+vars[[2]]*vars[[3]],type="l",col="red")
multitaper packageRead PDF manual