mwCor {astrochron}

Calculate moving window correlation coefficient for two stratigraphic series, using a 'dynamic window'

Description

Calculate moving window correlation coefficient for two stratigraphic series, using a 'dynamic window'. This routine adjusts the number of data points in the window so it has a constant duration in time or space, for use with unevenly sampled data.

Usage

mwCor(dat,cols=NULL,win=NULL,conv=1,cormethod=1,output=T,pl=1,genplot=T,verbose=T)

Arguments

dat

Your data frame containing stratigraphic data; any number of columns (variables) are permitted, but the first column should be a location identifier (e.g., depth, height, time).

cols

A vector that identifies the two variable columns to be extracted (first column automatically extracted).

win

Moving window size in units of space or time.

conv

Convention for window placement: (1) center each window on a stratigraphic level in 'dat' (DEFAULT), (2) start with the smallest location datum in 'dat', (3) start with the largest location datum in 'dat'. For options 2 and 3, the center of the window will not necessarily coincide with a measured stratigraphic level in 'dat', but edges of the data set are better preserved.

cormethod

Method used for calculation of correlation coefficient (1=Pearson, 2=Spearman, 3=Kendall)

output

Output results? (T or F)

pl

(1) Plot results at center of window, or (2) create "string of points plot" as in Sageman and Hollander (1999)

genplot

Generate summary plots? (T or F)

verbose

Verbose output? (T or F)

References

B.B. Sageman and D.H. Hollander, 1999, Cross correlation of paleoecological and geochemical proxies: A holistic approach to the study of past global change, in E. Barrera and C.C. Johnson, eds., GSA Special Paper 332, p. 365-384.

Examples

# generate example series
ex <- cycles(freqs=c(1/40,1/20),noisevar=.2)

# add second variable
ex[3] <- cycles(freqs=c(1/40,1/20),noisevar=0.2)[2]

# jitter sampling times
ex[1]=ex[1]+rnorm(500,sd=1)
# sort
ex = ex[order(ex[,1],na.last=NA,decreasing=FALSE),]

# run mwCor
mwCor(ex,win=50)