dplR (version 1.6.8)

strip.rwl: Chronology Stripping by EPS

Description

EPS-based chronology stripping after Fowler & Boswijk 2003.

Usage

strip.rwl(rwl, ids = NULL, verbose = FALSE, comp.plot = FALSE,
          legacy.eps = FALSE)

Arguments

rwl

a data.frame of raw tree-ring widths series, such as that produced by read.rwl or read.fh

ids

an optional data.frame with column one named "tree" giving a numeric ID for each tree and column two named "core" giving a numeric ID for each core. This is passed on to rwi.stats. See its manual for the meaning of the default value NULL and more information.

verbose

logical flag, indicating if the EPS calculated at each step and other details should be printed on the console during the chronology stripping process

comp.plot

logical flag, indicating if a diagnostic plot with year-wise stripped and unstripped EPS should be drawn (see details below)

legacy.eps

logical flag, indicating if the EPS will be calculated with rwi.stats (FALSE, the default) or rwi.stats.legacy (TRUE)

Value

The functions returns a data.frame of raw tree-ring widths, where series that do not contribute to an overall improvement in EPS are left out.

Details

The EPS-based chronology stripping is implemented after Fowler & Boswijk 2003: First, all series are standardized using a double detrending procedure with splines and frequency cutoffs of 50% at 20 and 200 years. Then, EPS is calculated for the chronology including all (remaining) series. In each iteration, the algorithm calculates leave-one-out EPS values, and the series whose removal increases overall EPS the most is discarded. This is repeated until no further increase in EPS is gained by discarding a single series. The procedure is then repeated in the opposite direction, i.e., the reinsertion of each previously removed series into the data.frame is considered. In each iteration, the series (if any) whose reinsertion increases EPS the most is reinserted. As a last step, EPS is calculated for each year of the stripped and original chronology including all series. If comp.plot is set to TRUE, a diagnostic plot is shown for the year-wise comparison.

When verbose output is chosen, the EPS values for all leave-one-out (or back-in) chronologies are reported. If discarding or re-inserting a single series leads to an improvement in EPS, this series is marked with an asterisk.

References

Fowler, A. and Boswijk, G. (2003) Chronology stripping as a tool for enhancing the statistical quality of tree-ring chronologies. Tree-Ring Research, 59(2), 53<U+2013>62.

See Also

rwi.stats

Examples

# NOT RUN {
library(utils)
data(anos1)
anos1.ids <- read.ids(anos1, stc = c(4, 3, 1))
srwl <- strip.rwl(anos1, ids = anos1.ids, verbose = TRUE)
tail(srwl)
# }