R/normalize1.R

In dplR: Dendrochronology Program Library in R

normalize1 <- function(rwl, n, prewhiten){
    rwl.mat <- as.matrix(rwl)
    ## Run hanning filter over the data if n isn't NULL
    ## divide by mean if n is null
    if(is.null(n)){
        master.stats <- colMeans(rwl.mat, na.rm=TRUE)
        master.mat <- sweep(rwl.mat, 2, master.stats, "/")
    } else {
        #master.stats <- apply(rwl.mat, 2, hanning, n)
        ## 15-dec-2022 AGB found a bug where a div0 was resulting in Nan. So recoding zeros.
        ## Apply hanning
        ## Recode any zero values to 0.001 -- here or after prewhiten? Same div0 bug should apply?
        master.stats <- apply(rwl.mat, 2, function(x){
          x2 <- hanning(x,n=n)
          x2[x2==0] <- 0.001
          x2
        })
        master.mat <- rwl.mat / master.stats
    }
    ## Apply ar if prewhiten
    if(prewhiten){
        ## take note of, ignore later, any columns without at least
        ## four observations
        idx.good <- colSums(!is.na(master.mat)) > 3
        master.mat <- apply(master.mat, 2, ar.func)
    } else {
        idx.good <- rep(TRUE, ncol(master.mat))
    }
    list(rwi.mat=master.mat, idx.good=idx.good)
}