dplR (version 1.6.8)

sea: Superposed Epoch Analysis

Description

This function calculates the significance of the departure from the mean for a given set of key event years and lagged years.

Usage

sea(x, key, lag = 5, resample = 1000)

Arguments

x

a chronology data.frame of ring-width indices (such as produced by chron)

key

a vector specifying the key event years for the superposed epoch

lag

an integral value defining the number of lagged years

resample

an integral value specifying the number of bootstrap sample for calculation of confidence intervals

Value

A data.frame with

lag

the lagged years,

se

the superposed epoch, i.e. the scaled mean RWI for the event years,

se.unscaled

the unscaled superposed epoch, i.e. the mean RWI for the event years,

p

significance of the departure from the chrono<U+2019>s mean RWI,

ci.95.lower

lower 95% confidence band,

ci.95.upper

upper 95% confidence band,

ci.99.lower

lower 99% confidence band,

ci.99.upper

upper 99% confidence band.

Details

Superposed epoch analysis (SEA) is used to test the significance of a mean tree growth response to certain events (such as droughts). Departures from the mean RWI values for the specified years prior to each event year, the event year, and the specified years immediately after each event are averaged to a superposed epoch. To determine if RWI for these years was significantly different from randomly selected sets of lag+1 other years, bootstrap resampling is used to randomly select sets of lag+1 years from the data set and to estimate significances for the departures from the mean RWI.

SEA computation is based on scaled RWI values, and 95%-confidence intervals are computed for the scaled values for each year in the superposed epoch.

References

Lough, J. M. and Fritts, H. C. (1987) An assessment of the possible effects of volcanic eruptions on North American climate using tree-ring data, 1602 to 1900 AD. Climatic Change, 10(3), 219<U+2013>239.

Examples

# NOT RUN {
library(graphics)
library(utils)
data(cana157)
event.years <- c(1631, 1742, 1845)
cana157.sea <- sea(cana157, event.years)
foo <- cana157.sea$se.unscaled
names(foo) <- cana157.sea$lag
barplot(foo, col = ifelse(cana157.sea$p < 0.05, "grey30", "grey75"), 
        ylab = "RWI", xlab = "Superposed Epoch")
# }