ClamR (version 2.1-3)

proxyJK: Jackknife Wilkinson Curve Fitting

Description

Perform a jackknife estimate of proxy curve fitting for time series analysis.

Usage

proxyJK(x, y, dx)

Value

List:

OUT

list( par, mid, ax, predmid,JKest, JKvar, PSTILDE )

omids

output midpoints

pmids

values at output midpoints

x

input x

y

input y

predy

predicted y from spline fit

Arguments

x

x-axis values

y

y-axis values

dx

width of window to span in time

Author

Jonathan M. Lees<[e-mail]>

Details

Routine that improves on the Wilkinson and Ivany(2002) approach to climate time series modeling. The jackknife is used to estimate the 95 percent confidence bounds for the modeled estimates. dx should be chosen to be approximately half a cycle or more.

References

Wang, T., Surge, D., and Lees, J. M., (2015) ClamR: A Statistical Evaluation of Isotopic and Temperature Records in Sclerochronologic Studies. Palaeogeography, Palaeoclimatology, Palaeoecology, doi:10.1016/j.palaeo.2015.07.008.

See Also

proxyA

Examples

if (FALSE) {

##########   this is for reading in data
########  fn = "/home/lees/DONNA/donna_viking_1.csv"

## fn = "donna_viking_1.csv"
######## C1 = scan(file=fn, what=list(mm="", o18=""), sep=",")
######## x = as.numeric(C1$mm)
######## y = as.numeric(C1$o18)
########x = x[!is.na(y)]
########y = y[!is.na(y)]


data(CLAM1)
x = CLAM1$x
y = CLAM1$y

dx = 3.392

gout = proxyJK(x, y, dx)

plotproxy1(x, y, gout)

par(mfrow=c(2,1))

plotproxy.error(x, y, gout, type = 1)
plotproxy.error(x, y, gout, type = 2)


par(mfrow=c(2,1))
plotproxy.error(x, y, gout, type = 2)

plotproxy.all2(gout, YAXstyle=1 )

}