Generate a log2 spaced EHA spectra — eha

Compute a log2 spaced Evolutive Harmonic Analysis (EHA) & Evolutive Power Spectral Analysis This is a wrapper function for the "eha" function of the 'astrochron' R package

Usage

eha_log2(
  data = NULL,
  win = NULL,
  tbw = NULL,
  demean = NULL,
  detrend = NULL,
  upperPeriod = NULL,
  lowerPeriod = NULL,
  pad = NULL,
  padding = "noise"
)

Arguments

data: Input series (matrix or data frame with two columns: depth/time in column 1, proxy value in column 2), preferably evenly sampled (the padded series is linearly interpolated with astrochron::linterp). Required (Default=NULL).
win: Length of the moving EHA window, in the depth/time unit of data (numeric, > 0, shorter than the record). The series is also padded by half a window at both ends (see padding). Required (Default=NULL).
tbw: Time-bandwidth product of the multitaper (MTM) spectra, passed to eha (numeric, <= 10, e.g. 2, 3 or 4); larger values give smoother spectra with a lower frequency resolution. Required (Default=NULL).
demean: Remove the mean of each window (logical). Currently not used: the windows are always demeaned (demean=TRUE is passed to eha). Default=NULL.
detrend: Remove a linear trend from each window (logical). Currently not used: the windows are always detrended (detrend=TRUE is passed to eha). Default=NULL.
upperPeriod: Longest period to analyse, in the depth/time unit of data (numeric, > lowerPeriod, at most about win); sets fmin = 1 / upperPeriod of the EHA and the top of the log2 period grid. Required (Default=NULL).
lowerPeriod: Shortest period to analyse, in the depth/time unit of data (numeric, > 0, at least twice the sampling interval); sets fmax = 1 / lowerPeriod and the bottom of the log2 period grid. Required (Default=NULL).
pad: Number of points to which each window is zero-padded (integer or NULL); odd values are increased by one, values below the number of points per window are raised to it, and 5 * pad is passed to eha to obtain a fine low-frequency spacing. NULL uses 2 * 2^ceiling(log2(points per window)). Default=NULL.
padding: Values used to extend the series by half a window at both ends, so that the EHA covers the whole record (character): "noise" (AR(1) red noise with the mean, standard deviation and lag-1 autocorrelation of the end segments; note that this calls set.seed()), "mean" (also "average" or "avg"; the mean of the series) or "zero" (zeros). Any other value is treated as "mean". Default="noise".

Value

The output is a list (analyze.eha object) which contain 12 objects which are the result of running the eha. Object 1: depth - depth/time of axis Object 2: log2_period - log2 scales period Object 3: pwr - Power values of the EHA run Object 4: amp - Amplitude values of the EHA run Object 5: prob - Probability values of the EHA run Object 6: f_test - F test values of the EHA run Scale size Object 7: Power.avg - Average power values Object 8: amp.avg - Average amplitude values Object 9: prob.avg - Average probability value Object 10: f_test.avg - Average f test values Object 11: x - x values of the data set Object 12: y - y values of the data set Object 13: window size

References

S.R. Meyers, 2012, Seeing Red in Cyclic Stratigraphy: Spectral Noise Estimation for #'Astrochronology: Paleoceanography, 27, PA3228, <doi:10.1029/2012PA002307>

S.R. Meyers, 2019 Cyclostratigraphy and the problem of astrochronologic testing, Earth-Science Reviews, Volume 190, <doi.org/10.1016/j.earscirev.2018.11.015.>

Author

Code based on on the "eha" function of the 'astrochron' R package

Examples

# \donttest{
#Example 1. Total Solar Irradiance
# data set of Steinhilber et al., (2012)

TSI_eha_log2 <- eha_log2(data = TSI,
win =  8192,
tbw = 4,
demean = TRUE,
detrend = TRUE,
upperPeriod = 8192,
lowerPeriod = 50,
pad = NULL,
padding = "noise")

#Example 2. Magnetic susceptibility data set of Pas et al., (2018)
mag_eha_log2 <- eha_log2(data = mag,
win =  50,
tbw = 4,
demean = TRUE,
detrend = TRUE,
upperPeriod = 50,
lowerPeriod = 2,
pad = NULL,
padding = "noise")

#Example 3. Greyscale data set of Zeeden et al., (2013)
grey_eha_log2 <- eha_log2(data = grey,
win =  20,
tbw = 4,
demean = TRUE,
detrend = TRUE,
upperPeriod = 20,
lowerPeriod = 2,
pad = NULL,
padding = "noise")

# }