Wrap spectral results for several datasets

This wrapper function is used to combine all main spectral results for the signal, noise and signal-to-noise ratio as presented and discussed in Münch and Laepple (2018), but it can also be used to combine the results for other data sets.

Usage

WrapSpectralResults(
  ...,
  diffusion = NULL,
  time.uncertainty = NULL,
  res = 1,
  df.log = 0.05,
  crit.diffusion = 0.5
)

Arguments

...: a comma separated list of named proxy datasets to analyse.
diffusion: a list the same length as the number of datasets with each list element a spectral object (?spec.object) of a transfer function to correct the corresponding dataset for the effect of diffusion-like smoothing (?SeparateSignalFromNoise for more details on this, and ?CalculateDiffusionTF for calculating transfer functions specifically for the firn diffusion process). Internally, the inverse of the transfer function values are applied to correct for the smoothing effect on the estimated signal and noise spectra (see Eq. 4 in Münch and Laepple, 2018). If NULL, no correction is applied at all; if instead you want to omit the correction only for some specific data set(s), set the corresponding list element(s) to NA.
time.uncertainty: as diffusion a list of transfer functions to correct for the effect of time uncertainty (?CalculateTimeUncertaintyTF for calculating transfer functions in the case of layer-counted proxy chronologies).
res: the sampling (e.g., temporal) resolution of the proxy data. Either a single value if all datasets have the same resolution, or a vector with a value for each dataset.
df.log: a vector of Gaussian kernel widths in log space to smooth the spectral estimates from each dataset. Either a single value to apply the same smoothing to all datasets, or a vector with a value for each dataset.
crit.diffusion: minimum transfer function value for the diffusion-like smoothing process to constrain the corresponding correction. This determines a cutoff frequency until which results are analysed to avoid large uncertainties at the high-frequency end of the spectra; defaults to 0.5.

Value

A list of n lists, where n is the number of provided datasets and where each of these lists contains up to four elements:

raw:: a list with four elements: three spectral objects (the raw signal, noise and corresponding SNR spectra), and a two-element vector (f.cutoff) with the index and value of the cutoff frequency from constraining the smoothing correction (see the crit.diffusion parameter).
corr.diff.only:: as item raw but with the spectra after correction for the effect of diffusion-like smoothing.
corr.t.unc.only:: as item raw but with the spectra after correction for the effect of time uncertainty.
corr.full:: as item raw but with the spectra after correction for both the effects of diffusion-like smoothing and time uncertainty.

The number of the returned list elements for each dataset depends on whether transfer functions for the corrections have been provided in diffusion and time.uncertainty or not. Also, the element f.cutoff is NA if diffusion-like smoothing has not been corrected for.

References

Münch, T. and Laepple, T.: What climate signal is contained in decadal- to centennial-scale isotope variations from Antarctic ice cores? Clim. Past, 14, 2053–2070, https://doi.org/10.5194/cp-14-2053-2018, 2018.

Author

Thomas Münch

Examples

# Get main results of Münch and Laepple (2018)

results <- WrapSpectralResults(
               dml1 = dml$dml1, dml2 = dml$dml2, wais = wais,
               diffusion = diffusion.tf,
               time.uncertainty = time.uncertainty.tf,
               df.log = c(0.15, 0.15, 0.1))