CFtime

CFtime is an R package that supports working with CF Metadata Conventions time coordinates, specifically geared to time-referencing data sets of climate projections such as those produced under the World Climate Research Programme and re-analysis data such as ERA5 from the European Centre for Medium-range Weather Forecasts (ECMWF).

The data sets include in their metadata an epoch, or origin, a point in time from which other points in time are calculated. This epoch takes the form of days since 1949-12-01, with each data collection (Coupled Model Intercomparison Project (CMIP) generation, model, etc) having its own epoch. The data itself has a temporal dimension if a coordinate variable in the netCDF file has an attribute units with a string value describing an epoch. The coordinate variable, say “time”, has data values such as 43289, which are offsets from the epoch in units of the epoch string (“days” in this case). To convert this offset to a date, using a specific calendar, is what this package does. Given that the calendars supported by the CF Metadata Conventions are not compatible with POSIXt, this conversion is not trivial because the standard R date-time operations do not give correct results. That it is important to account for these differences is easily demonstrated:

library(CFtime)

# POSIXt calculations on a standard calendar
as.Date("1949-12-01") + 43289
#> [1] "2068-06-08"

# CFtime calculation on a "360_day" calendar
as_timestamp(CFtime("days since 1949-12-01", "360_day", 43289))
#> [1] "2070-02-30"

That’s a difference of nearly 21 months! (And yes, 30 February is a valid date on a 360_day calendar.)

All defined calendars of the CF Metadata Conventions are supported:

• standard or gregorian: This calendar is valid for the Common Era only; it starts at 0001-01-01 00:00:00, i.e. 1 January of year 1. Time periods prior to the introduction of the Gregorian calendar (1582-10-15) use the julian calendar that was in common use then. The 10-day gap between the Julian and Gregorian calendars is observed, so dates in the range 5 to 14 October 1582 are invalid.
• proleptic_gregorian: This calendar uses the Gregorian calendar for periods prior to the introduction of that calendar as well, and it extends to periods before the Common Era, e.g. year 0 and negative years.
• tai: International Atomic Time, a global standard for linear time based on multiple atomic clocks: it counts seconds since its start at 1958-01-01 00:00:00. For presentation it uses the Gregorian calendar. Timestamps prior to its start are not allowed.
• utc: Coordinated Universal Time, the standard for civil timekeeping all over the world. It is based on International Atomic Time but it uses occasional leap seconds to remain synchronous with Earth’s rotation around the Sun; at the end of 2024 it is 37 seconds behind tai. It uses the Gregorian calendar with a start at 1972-01-01 00:00:00; earlier timestamps are not allowed. Future timestamps are also not allowed because the insertion of leap seconds is unpredictable. Most computer clocks synchronize against UTC but calculations of periods do not consider leap seconds.
• julian: The julian calendar has a leap year every four years, including centennial years. Otherwise it is the same as the standard calendar.
• 365_day or noleap: This is a “model time” calendar in which no leap years occur. Year 0 exists, as well as years prior to that.
• 366_day or all_leap: This is a “model time” calendar in which all years are leap years. Year 0 exists, as well as years prior to that.
• 360_day: This is a “model time” calendar in which every year has 360 days divided over 12 months of 30 days each. Year 0 exists, as well as years prior to that.
• none: Perpetual “calendar” for experiments that are simulated on a given instant during the year. All the elements in this calendar thus represent the same instant in time.

Use of custom calendars is currently not supported.

This package facilitates use of a suite of models of climate projections that use different calendars in a consistent manner. This package is particularly useful for working with climate projection data having a daily or higher resolution, but it will work equally well on data with a lower resolution.

Timestamps are generated using the ISO8601 standard.

Calendar-aware factors can be generated to support processing of data using tapply() and similar functions. Merging of multiple data sets and subsetting facilitate analysis while preserving computer resources.

Installation

Get the latest stable version on CRAN:

install.packages("CFtime")

You can install the development version of CFtime from GitHub with:

# install.packages("devtools")
devtools::install_github("pvanlaake/CFtime")

Basic operation

The package contains a class, CFTime, to describe the time coordinate reference system, including its calendar and origin, and which holds the time coordinate values that are offset from the origin to represent instants in time. This class operates on the data in the file of interest, here a Coordinated Regional Climate Downscaling Experiment (CORDEX) file of precipitation for the Central America domain.

In this vignette we are using the ncdfCF package as that provides the easiest interface to work with netCDF files. Package CFtime is integrated into ncdfCF which makes working with time dimensions in netCDF seamless.
Packages RNetCDF and ncdf4 can work with CFtime as well but then the “intelligence” built into ncdfCF is not available, such as automatically identifying axes and data orientation. Other packages like terra and stars are not recommended because they do not provide access to the specifics of the time dimension of the data and do not properly consider any calendars other than “standard” and “proleptic_gregorian”.

# install.packages("ncdfCF")
library(ncdfCF)

# Opening a data set that is included with the package.
# Usually you would `list.files()` on a directory of your choice.
fn <- list.files(path = system.file("extdata", package = "CFtime"), full.names = TRUE)[1]
ds <- open_ncdf(fn)
ds$attribute("title")
#> [1] "NOAA GFDL GFDL-ESM4 model output prepared for CMIP6 update of RCP4.5 based on SSP2"
ds$attribute("license")
#> [1] "CMIP6 model data produced by NOAA-GFDL is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License (https://creativecommons.org/licenses/). Consult https://pcmdi.llnl.gov/CMIP6/TermsOfUse for terms of use governing CMIP6 output, including citation requirements and proper acknowledgment. Further information about this data, including some limitations, can be found via the further_info_url (recorded as a global attribute in this file). The data producers and data providers make no warranty, either express or implied, including, but not limited to, warranties of merchantability and fitness for a particular purpose. All liabilities arising from the supply of the information (including any liability arising in negligence) are excluded to the fullest extent permitted by law."

# What axes are there in the data set?
dimnames(ds)
#> [1] "bnds" "lat"  "time" "lon"

# Get the CFTime instance from the "time" axis
(time <- ds[["time"]]$time())
#> CF calendar:
#>   Origin  : 1850-01-01T00:00:00
#>   Units   : days
#>   Type    : noleap
#> Time series:
#>   Elements: [2015-01-01T12:00:00 .. 2099-12-31T12:00:00] (average of 1.000000 days between 31025 elements)
#>   Bounds  : irregular

Note that the ncdfCF package reads the netCDF file and interprets its contents on the basis of its attribute values. If an axis is found that represents time, then a CFTime instance is created for it, which can be accessed with the time() method.

Using RNetCDF or ncdf4

If you are using the RNetCDF or ncdf4 package rather than ncdfCF, creating a CFTime instance goes like this (but note that this assumes that the axis is called “time”):

library(RNetCDF)
nc <- open.nc(fn)
time <- CFtime(att.get.nc(nc, "time", "units"), 
               att.get.nc(nc, "time", "calendar"), 
               var.get.nc(nc, "time"))

library(ncdf4)
nc <- nc_open(fn)
names(nc$var) # A mix of data variables, axes, and other objects
t <- CFtime(nc$dim$time$units, 
            nc$dim$time$calendar, 
            nc$dim$time$vals)

Typical workflow

In a typical process, you would combine multiple data files into a single data set to analyze a feature of interest. To continue the previous example of precipitation in the Central America domain using CORDEX data, you can calculate the precipitation per month for the period 2041 - 2050 as follows:

# NOT RUN
library(ncdfCF)
library(abind)

# Open the files - one would typically do this in a loop
ds2041 <- open_ncdf("~/pr_CAM-22_MOHC-HadGEM2-ES_rcp26_r1i1p1_GERICS-REMO2015_v1_day_20410101-20451230.nc")
ds2046 <- open_ncdf("~/pr_CAM-22_MOHC-HadGEM2-ES_rcp26_r1i1p1_GERICS-REMO2015_v1_day_20460101-20501230.nc")

# Create the time object from the first file
# All files have an identical "time" axis as per the CORDEX specifications
time <- ds2041[["time"]]$time()

# Add the time values from the remaining files
time <- time + ds2046[["pr"]]$time()$offsets

# Grab the data from the files and merge the arrays into one, in the same order
# as the time values
pr <- abind(ds2041[["pr"]]$data()$array(), ds2046[["pr"]]$data()$array())

# Create the month factor from the time object
f_month <- CFfactor(time, "month")

# The result from applying this factor to a data set that it describes is a new
# data set with a different "time" dimension. The function result stores this
# new time object as an attribute.
pr_month_time <- attr(f_month, "CFTime")

# Now sum the daily data to monthly data
# Dimensions 1 and 2 are longitude and latitude, the third dimension is time
pr_month <- aperm(apply(pr, 1:2, tapply, f_month, sum), c(2, 3, 1))
dimnames(pr_month)[[3]] <- as_timestamp(pr_month_time)

Coverage

This package has been tested with the following data sets:

• ERA5 (including multiple variables, levels, and mixed ERA5/ERA5T data)
• CMIP5
• CORDEX
• CMIP6
• ROMS

The package also operates on geographical and/or temporal subsets of data sets so long as the subsetted data complies with the CF Metadata Conventions. This includes subsetting in the Climate Data Store. Subsetted data from Climate4Impact is not automatically supported because the dimension names are not compliant with the CF Metadata Conventions, use the corresponding dimension names instead.