In nickmckay/lipdR: LiPD utilities for R

options(width = 500)
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Quick summaries for LiPD objects

LiPD datasets can contain a lot of data and metadata, and it's often useful to get a quick summary of one or more datasets that you've loaded into R. Let's load in a file from lipdverse.org and see how this works.

LiPD objects

Let's first grab at a single LiPD the Moose Lake dataset from Clegg et al. (2010) that's part of the PAGES2k Temperature compilation.

library(lipdR)
L <- readLipd("https://lipdverse.org/data/MD6jkgwSxsq0oilgYUjM/1_0_0//Arc-MooseLake.Clegg.2010-ensemble.lpd")

First, let's get a quick look using the print() function, or by just returning the object:

L

This is handy, as we can quickly see...

To get more detail about the dataset, let's use summary()

summary(L)

In addition to the information printed by print(), summary() includes tables with basic metadata about each of the columns in the measurement tables. In the Moose Lake dataset's paleoData measurement table, we see two variables, temperature and year, and the table shows the units, description, and some stats about the contents. Similarly, we get information about the radiocarbon dates included in the chronData measurement table.

Multi LiPD objects

LiPD datasets are often most powerful when many datasets are analyzed together, and it's common to load multiple datasets into a "multi_lipd" object. It's nice to get useful summaries for these data too. Let's explore this functionality using a diverse collection of LiPD files pulled at random from around the LiPDverse.

D <- readLipd("http://lipdverse.org/testData/testData.zip")

Once again, we can print() for an overview:

D

and see the number of datasets included, their archiveTypes and geographic boundaries.

We can also have a look at any individual dataset in the collection by subsetting the multi-lipd:

D$`Arc-Agassiz.Vinther.2008`

For more information on the multi-lipd, we can use summary()

summary(D)

And also get tables that give us some basic information on each of the datasets included. There are a few options that customize the output of summary, here we will choose the preferred time units, and limit the number of datasets summarized to 20:

summary(D, print.length=20, time.units="AD")

These options limit the printed information on the table to 20 rows, and gives us time summary information in AD units where possible.

Storing the summary table

Sometimes it is useful to save the table, so it can be sorted and further investigated. Simply assign an output:"

MLSummary <- summary(D)

We can now explore this table as a standard data frame. Here we do so using dplyr tools to sort by datasets with age ensembles

library(dplyr)
arrange(MLSummary,desc(NumEns))

With a well curated dataset (this example doesn't quite fit those criteria), a few commands can get you pretty close to a publishable table.

*NPM: we might have to kill this section because of the gt package, or just not show that. *

library(gt)
MLSummary %>% 
  select(-starts_with("Num"),-"Paleo Vars") %>% 
  gt()

LiPD Timeseries objects

Typically, when we load multiple LiPD datasets, it's convenient to convert them into timeseries objects, as it greatly simplifies analyzing the data. There are print and summary methods for these objects too. There are a couple different representations of these objects availabe in R - namely, as a list (lipd_ts) or as a nested tibble (lipd_ts_tibble). If you're used to working with tibbles and tidyverse you'll probably prefer the latter, so we'll do that here:

tibTS <- as.lipdTsTibble(D)

Let's quickly check the size of this tibble

dim(tibTS)

Wow - that's pretty big. That means that there are r nrow(tibTS) variables across all our datasets, and r ncol(tibTS) data or metadata fields present in one or more of the datasets.

Again, let's print out a quick overview:

tibTS

and we see the number of datasets, variables, and data/metadata fields. We also get the time units of all variables and the time interval common to all variables. These data have not been curated, and we see that no time units or time interval is common to all variables.

Beyond this quick overview, we can get a lot more information using summary(). This works the same for lipd_ts and lipd_ts_tibble objects. Let's limit the table to show details of the first 10 rows only

summary(tibTS, print.length = 10)

In addition to the overview provided by print(), summary() gives us a table that shows us which variables are most common in the TS object. We can explore this more by specifying which metadata fields we'd like to summarize using the add.variable parameter.

summary(tibTS,
        print.length = 10, 
        add.variable = c("archiveType", "paleoData_meanValue12k","geo_continent", "geo_elevation", "agesPerKyr", "dataSetName"))

Now we get the same summarizing information for those additional six columns. The character or categorical data (archiveType, geo_contintent, and dataSetName) get added to the count table. The numeric data (paleoData_meanValue12k, geo_elevation, and agesPerKyr) are summarized in quantiles above.

Filtering the TS object

This summary information is often a great first step towards informing how you will filter or subset your TS object for future analysis. For example, let's grab only the variables that:

1. are in the continental US (roughly)
2. cover the early Holocene
3. Have some climate interpretation

library(magrittr) #We'll use the pipe for clarity
subsetTs <- tibTS %>% 
  filter(between(geo_latitude,30,50),
         between(geo_longitude,-125,-70),
         maxYear > 11700,
         minYear < 8000,
         !is.na(interpretation1_variable))

Great - we found r nrow(subsetTs) timeseries that meet these criteria. Let's explore their interpretations in a bit more detail by saving the table output.

sumTable <- summary(subsetTs,
                    add.variable = c("interpretation1_variable",
                                     "interpretation1_variableDetail",
                                     "interpretation1_seasonality"))

nickmckay/lipdR documentation built on April 13, 2025, 5:58 p.m.