dplR (version 1.7.6)

pass.filt: Low-pass, high-pass, band-pass, and stop-pass filtering

Description

Applies low-pass, high-pass, band-pass, or stop-pass filtering to y with frequencies (or periods) supplied by the user.

Usage

pass.filt(y, W, type = c("low", "high", "stop", "pass"),
          method = c("Butterworth", "ChebyshevI"),
          n = 4, Rp = 1)

Value

A filtered vector.

Arguments

y

a numeric vector, typically a tree-ring series.

W

a numeric vector giving frequency or period of the filter. See details.

type

a character giving the type of filter. Values can be "low", "high", "stop", or "pass" for low-pass, high-pass, band-pass, or stop-pass filters. Defaults to "low".

method

a character specifying indicating whether to use a Butterworth (default) or a type I Chebyshev filter.

n

a numeric value giving the order of the filter. Larger numbers create steeper fall off.

Rp

a numeric value giving the dB for the passband ripple.

Author

Andy Bunn. Patched and improved by Mikko Korpela.

Details

This function applies either a Butterworth or a Chebyshev type I filter of order n to a signal and is nothing more than a wrapper for functions in the signal package. The filters are dsigned via butter and cheby1. The filter is applied via filtfilt.

The input data (y) has the mean value subtracted and is then padded via reflection at the start and the end to a distance of twice the maximum period. The padded data and the filter are passed to filtfilt after which the data are unpadded and returned afer the mean is added back.

The argumement W can be given in either frequency between 0 and 0.5 or, for convenience, period (minimum value of 2). For low-pass and high-pass filters, W must have a length of one. For low-pass and high-pass filters W must be a two-element vector (c(low, high)) specifying the lower and upper boundaries of the filter.

Because this is just a wrapper for casual use with tree-ring data the frequencies and periods assume a sampling frequency of one. Users are encouraged to build their own filters using the signal package.

See Also

hanning, detrend

Examples

data("co021")
x <- na.omit(co021[, 1])

# 20-year low-pass filter -- note freq is passed in
bSm <- pass.filt(x, W=0.05, type="low", method="Butterworth")
cSm <- pass.filt(x, W=0.05, type="low", method="ChebyshevI")
plot(x, type="l", col="grey")
lines(bSm, col="red")
lines(cSm, col="blue")

# 20-year high-pass filter -- note period is passed in
bSm <- pass.filt(x, W=20, type="high")
plot(x, type="l", col="grey")
lines(bSm, col="red")

# 20 to 100-year band-pass filter -- note freqs are passed in
bSm <- pass.filt(x, W=c(0.01, 0.05), type="pass")
cSm <- pass.filt(x, W=c(0.01, 0.05), type="pass", method="ChebyshevI")
plot(x, type="l", col="grey")
lines(bSm, col="red")
lines(cSm, col="blue")

# 20 to 100-year stop-pass filter -- note periods are passed in
cSm <- pass.filt(x, W=c(20, 100), type="stop", method="ChebyshevI")
plot(x, type="l", col="grey")
lines(cSm, col="red")