Smooth and gap-fill a time series using Whittaker smoothing

Applies weighted Whittaker smoothing to a numeric time series, filling short gaps (NAs) and smoothing the signal at the same time. This function is useful for environmental time series with missing or noisy data. Note that the input time series need to be sampled at a regular interval. If not, you need to resample it first, with NAs inserted at the missing time points.

Usage

whittaker_smoothing_filling(x, lambda, maxgap = Inf, minseg = 2)

Arguments

x

Numeric vector. The time series signal to be smoothed (e.g., EVI, NDVI, or other index).

lambda

Numeric. Smoothing parameter; larger values result in a smoother output.

maxgap

Numeric. Maximum number of consecutive NAs to interpolate (default: Inf). Gaps longer than this will remain NA.

minseg

Numeric. Minimum segment length for smoothing (default: 2). Segments shorter than this will be replaced with NA.

Value

Numeric vector. The smoothed and gap-filled signal, with the same length as x.

Examples

if (FALSE) { # \dontrun{
# Simulate a noisy, gappy time series
set.seed(42)
t <- 1:365
x <- sin(2 * pi * t / 365) + rnorm(365, sd = 0.1)
x[sample(1:365, 30)] <- NA

# Smooth and fill gaps
x_sm <- whittaker_smoothing_filling(x, lambda = 50, maxgap = 14, minseg = 2)
plot(t, x, type = "p", col = "grey", main = "Whittaker Smoothing")
lines(t, x_sm, col = "darkgreen", lwd = 2)
} # }