In paleovar/ptboxproxytools: PTBox proxy tools for time series processing and analysis

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Applications of package functions to example data

Loading requirements

library(PTBoxProxytools) # remove for public
library(PTBoxProxydata) # loads PTBoxProxydata dependency
library(magrittr) # for the %>% operator

Applying functions to different PTBoxProxydata time series objects

First, we create a ProxyDataManager instance and load the icecore_testset built into PTBoxProxydata as data for testing the PTBoxProxytools functions. This outputs a PTBoxProxydata::Proxytibble object. This will be used to explain how to apply the package functions to it.

mng <- ProxyDataManager()

icecore_data <- load_set(mng, dataset_names = 'icecore_testset', zoo_format = 'zoo') %>% clean_timeseries(.)
print(icecore_data)

Please find also the vignette("PTBoxProxydata_howto") in the PTBoxProxydata package for an introduction to the data objects provided by PTBoxProxydata.

In a second step, we explore the basic usage for PTBoxProxytools functions that operate on proxy data. Therefore, we apply PTBoxProxytools::paleodata_windowing as an exemplary function to an individual time series (zoo::zoo)

# 1. example: apply `Proxytools` function (here `paleodata_windowing`)
# to an individual `zoo::zoo`
testout <- icecore_data$proxy_data[[1]] %>% 
  paleodata_windowing(.,4000,6000)
plot(testout)

and to all time series stored in a Proxytibble (PTBoxProxydata::Proxytibble).

# 2. example: apply function or to all `zoo::zoo`s/`PTBoxProxydata::Proxyzoo`s
# contained in a `PTBoxProxydata::Proxytibble`
testout <- icecore_data %>% 
  paleodata_windowing(.,4000,6000)
print(testout)

Some useful examples of PTBoxProxytools functions

1. Normalization to center and rescale irregular time series

# normalization
testout <- icecore_data$proxy_data[[1]] %>% 
  normalize()
head(testout,3)

2. Interpolation for irregular time series

# interpolation
# internally uses PTBoxProxydata::zoo_apply
some_window <- icecore_data$proxy_data[[1]] %>% 
  zoo::index() %>%
  .[c(1,length(.))]
outtimes <- seq(some_window[1], some_window[2], length.out = 20)
testout <- icecore_data$proxy_data[[1]] %>% 
  paleodata_interpolation(., "spectral", outtimes)
head(testout,3)
print(dim(testout)[1])
# output is still a zoo
print(class(testout))

3. Gaussian filtering of irregular time series

# Detrend ice core data with 10kyr cutoff timescale
icecoredata_detrended <- paleodata_filtering(icecore_data, 'detrend', detr_scale=10000)
# Smooth the data with 10kyr cutoff timescale
icecoredata_smoothed <- paleodata_filtering(icecore_data, 'smooth', smooth_scale=10000)
# Apply bandpass filter for timescales from 1kyr to 10kyr
icecoredata_filtered <- paleodata_filtering(icecore_data, 'bandpass', filter_scales=data.frame(lower=1000,upper=10000))
# Plot results
#plot(icecoredata_detrended$proxy_data[[1]])
#plot(icecoredata_smoothed$proxy_data[[1]])
plot(icecoredata_filtered$proxy_data[[1]]) #we show only one example plot
plot(icecoredata_filtered$proxy_data[[2]]) #we show only one example plot

4. Spectral analysis

First, we compute and plot the spectrum.

testout <- icecore_data$proxy_data[[1]] %>%
  paleodata_spectrum(.)

library(PaleoSpec)
PaleoSpec::LPlot(testout$EDCBag_18O$raw, col="grey")
PaleoSpec::LLines(testout$EDCBag_18O$logsmooth, col="darkblue")

Second, we compute the variance on timescales of 200 to 2000 years, which is stored in the proxy_data column.

testout <- icecore_data %>%
  paleodata_varfromspec(.,  target="raw", freq.start = 5e-4, freq.end = 5e-3)
head(testout$proxy_data)

Third, we fit a power-law scaling to the spectrum on timescales of 2000 to 20000 years, which is stored again in proxy_data.

testout <- icecore_data %>%
  paleodata_scaling(.,  target="raw", freq.start = 5e-5, freq.end = 5e-4)
head(testout$proxy_data,3)

paleovar/ptboxproxytools documentation built on June 9, 2025, 1:40 a.m.