Adaptive eCOCO

Adaptive COCO in physical-depth sliding windows with per-window AR(1) Monte Carlo nulls and ridge tracking (Acycle ecoco.m adaptive branch, ecocoAdaptiveCore.m; Mingsong Li).

Usage

ecoco_adaptive(
  data,
  orbit9 = orbit9_default(),
  window,
  sr = seq(0.5, 20, by = 0.1),
  step = 1,
  step_depth = NULL,
  pad = 5000,
  red = 1,
  nsim = 500,
  method = c("Pearson", "Spearman"),
  max_frequency = NULL,
  amplitude_mode = c("four-group-area", "per-orbit", "fixed"),
  seed = 1,
  center_limits = NULL,
  batch = NULL,
  verbose = TRUE,
  centre_limits = NULL
)

Arguments

data

Record to analyse (matrix or data frame, two columns): depth in metres in column 1 and the proxy in column 2. Non-finite rows are dropped, duplicated depths averaged, and unevenly spaced data are linearly interpolated to the median sample spacing (see coco_prepare_regular_data). Required, no default.

orbit9

The nine astronomical target periods in kyr (numeric vector of nine distinct values: long eccentricity, four short eccentricity terms, obliquity, three precession terms), e.g. from orbit9_at_age or calculate_orbit9 for the age of the record. The default uses the present-day periods (405.7, 130.7, 123.9, 98.9, 94.9, 41.0, 23.7, 22.4 and 19.0 kyr). Default=orbit9_default().

window

Length of each sliding window, in metres (numeric, > 0, shorter than the record; every window must contain at least 4 data points). Each window is resampled to an odd number of points around its centre and analysed with Adaptive COCO. Required, no default.

sr

Sedimentation rates to test in every window, in cm/kyr (numeric vector, > 0). Default=seq(0.5, 20, by = 0.1).

step

Step between successive window centres, in samples of the (regularised) record (numeric, > 0); only used when step_depth=NULL, giving step_depth = step * sample interval. Default=1.

step_depth

Exact spacing of the window centres, in metres (numeric, > 0, or NULL); overrides step. At most 10000 windows are allowed. Default=NULL.

pad

Length (number of points) to which every window is zero-padded for the periodogram (integer, at least the number of points per window); a larger value gives a finer frequency grid. Default=5000.

red

Background (red-noise) removal applied to every window spectrum before the correlation (integer): 0 no removal, 1 classical AR(1) background subtracted, 2 robust AR(1) background (median-smoothed spectrum, Mann and Lees 1996 style), 3 smoothed-window-average (SWA) background (needs at least 33 spectral bins). Negative power after subtraction is set to zero. Default=1.

nsim

Number of AR(1) Monte Carlo simulations per window used for the global and local p-values (integer, >= 0); the AR(1) coefficient and standard deviation are estimated in each window. 0 skips the simulations, so the p-values are NA. Default=500.

method

Correlation coefficient between the window spectrum and the target spectrum (character): "Pearson" or "Spearman". Default=c("Pearson", "Spearman"), i.e. "Pearson".

max_frequency

Highest frequency of the target spectrum that enters the correlation, in cycles/kyr (numeric, must be at least the highest orbital frequency max(1/orbit9)); NULL uses 1.2 * max(1/orbit9). Passed to coco_adaptive_evaluate. Default=NULL.

amplitude_mode

How the amplitudes of the nine target sinusoids are set (character): "four-group-area" fits one amplitude per group (long eccentricity, short eccentricity, obliquity, precession) to the band areas of the data spectrum (Acycle default), "per-orbit" fits each of the nine amplitudes to the spectral peak at its frequency, "fixed" uses fixed weights (1.0 eccentricity, 0.8 obliquity, 0.6 precession; see coco_fixed_target_weights). The target is always synthesised as "coherent-nine". Default=c("four-group-area", "per-orbit", "fixed"), i.e. "four-group-area".

seed

Seed of the random number generator (integer), so that the Monte Carlo results can be reproduced; NULL leaves the current random state unchanged. Default=1.

center_limits

Depths of the first and last window centre, in metres (numeric vector c(first, last), or NULL); the windows must still fit inside the record. NULL uses c(first depth + window/2, last depth - window/2). Default=NULL.

batch

Number of Monte Carlo simulations processed together in one batch (integer, >= 1); larger batches are faster but use more memory. NULL chooses min(50, nsim, 3e6 / (frequencies * windows)). Default=NULL.

verbose

Print messages about the windows and the progress of the Monte Carlo simulations (logical, TRUE or FALSE). Default=TRUE.

centre_limits

British spelling of center_limits; if given (not NULL) it overrides center_limits. Default=NULL.

Value

list with matrices (rate-by-window): rho, p_global, p_local, norbit, pcoco, score; `track` = ridge-tracked sedimentation-rate path