The neonUtilities R package provides utilities for discovering, downloading, and working with NEON data files. NEON data files can be downloaded from the NEON Data Portal (http://data.neonscience.org) or API (http://data.neonscience.org/data-api). Provisional NEON data files from instrumented and observation systems are delivered by NEON within zip files organized by site and year-month. Provisional NEON data files from the airborne observation platform (AOP) are organized by site and year.
neonUtilities is available on CRAN and most users will want to install it from there. If you want to use the current development version, you can install from GitHub, but be warned that the version here may not be stable.
This package was developed on top of the deprecated neonDataStackR package; change logs from that package are included below.
This package is under development - please post any issues here and tag @chrlaney and/or @cklunch.
To get citation details for citing the neonUtilities package in a publication, run citation("neonUtilities") in R.
Install the package into your local environment using the following code:
install.packages('neonUtilities')
library(neonUtilities)zipsByProduct() Download NEON observational (OS) and instrumentation (IS) data by data product.byFileAOP() Download NEON remote sensing data by site, year, and data product.byTileAOP() Download NEON remote sensing data by data product, year, and UTM coordinates.stackByTable() NEON OS and IS data are provided in separate files for each site and month of collection (and location within site, for IS). This function merges these data files cleanly for each data type. Works with data downloaded either via the NEON Data Portal or zipsByProduct().readTableNEON() Read NEON tabular data into R, detecting variable types via the variables file. Currently works for OS and IS data, but not eddy covariance data.loadByProduct() Combines the actions of zipsByProduct(), stackByTable(), and readTableNEON() for OS and IS data. In one step, downloads, merges, and loads data into R.stackEddy() Extracts NEON eddy covariance data from the provided HDF files and merges the resulting tabular data, similar to the IS data handling in stackByTable().footRaster() Makes a raster of the flux footprint data provided in the expanded data package for eddy covaraince.See NEON Data Tutorials for more information, particularly the Download and Explore and neonUtilities tutorials.
stackByTable() unzips monthly packages, finds the CSV data files, and joins them by table (e.g., 2DWSD_2min, 2DWSD_30min for 2D Wind Speed and Direction). For data products from instrumented systems that have multiple sensors placed at various heights (or depths) and/or horizontal positions away from the supporting tower, this function will create 2 columns in addition to the existing columns, one for horizontalPosition and the other for verticalPosition. This function will only work for data products that organize data in CSV files. Other data file types, such as HDF5 files from the eddy covariance system and remote sensing airborne observing platform (AOP) are not supported.
stackByTable(filepath = "testdata/NEON_size-dust-particulate.zip") # modify filepath to your directoryTo load data directly into the current R environment, instead of saving the stacked files to the filepath, use the option savepath='envt'. When using this option, assign the output of the function to a variable name. The output object will be a named list of data tables.
dust <- stackByTable(filepath = "testdata/NEON_size-dust-particulate.zip", savepath="envt")getPackage() can be used to pull a single zip file (all the data for a single data product by site by month combination) using the NEON API.
# Plant phenology observations from the Jornada LTER site, May 2017
getPackage(dpID = "DP1.10055.001", site_code = "JORN", year_month = "2017-05", package = "basic")zipsByProduct() pulls data from the NEON API in the correct format to be stacked by stackByTable(). Depending on the data product and data volume, pulling data from the API with zipsByProduct() can take a very long time.
{
# Herbaceous clip harvest data, from all sites and months for which it is currently available
zipsByProduct(dpID="DP1.10023.001", site="all", package="basic", check.size=T)
stackByTable(paste0(getwd(), "/filesToStack10023"))
}loadByProduct() performs the actions of both zipsByProduct() and stackByTable() and loads the resulting data into the current R environment. The object output by loadByProduct() is a named list of tables.
bird <- loadByProduct(dpID="DP1.10003.001", site="all", package="expanded")
names(bird)
# To get each table in the list as an independent object, outside of the list:
list2env(bird, .GlobalEnv)Both zipsByProduct() and loadByProduct() can also subset by sites and date range:
wq <- loadByProduct(dpID="DP1.20288.001", site=c("ARIK","POSE"), startdate="2018-04", enddate="2018-08")byFileAOP() pulls data from the NEON API, specifically for remote sensing (AOP) data. This function preserves the file directory hierarchy that AOP files are typically stored in, making it easier to navigate a large number of downloaded files.
# Lidar slant rangeform data from Santa Rita Experimental Range, 2017 flight
byFileAOP(dpID = "DP3.30001.001", site = "SRER", year = "2017", check.size = T)byTileAOP() pulls AOP data from the NEON API, for tiles matching the coordinates specified in the function call. A buffer can also be included in the function call, to download tiles within a certain distance of the coordinates (e.g., if the coordinates are the plot centroids of NEON TOS plots, use buffer=20). byTileAOP() will only work on the Level 3, mosaicked AOP products.
# Vegetation indices from San Joaquin Experimental Range, 2017
# easting and northing must be matched vectors of UTM coordinates
byTileAOP(dpID="DP3.30026.001", site="SJER", year="2017", easting=easting, northing=northing, buffer=20)transformFileToGeoCSV() takes any single NEON csv data file plus its respective variables file, and generates a new CSV with GeoCSV headers. This makes the data similar in format to data provided by organizations such as UNAVCO, and is good for embedding in a repeating script.
For a tutorial explaining how to use the neonUtilities package in more detail, including additional input options, view the Use the neonUtilities Package to Access NEON Data tutorial.
A cheat sheet for the neonUtilities package is available here.
zipsByProduct() and byFileAOP() use the download.file() function, wrapped by the downloader package, and we’ve found in testing that download.file() can be finicky. Using R version > 3.4 seems to help, and if you’re on Windows, using Windows 10. Feel free to contact us if you run into problems!The National Ecological Observatory Network is a project solely funded by the National Science Foundation and managed under cooperative agreement by Battelle. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
GNU AFFERO GENERAL PUBLIC LICENSE Version 3, 19 November 2007
Information and documents contained within this repository are available as-is. Codes or documents, or their use, may not be supported or maintained under any program or service and may not be compatible with data currently available from the NEON Data Portal.
Enhancements: * getNeonDOI() function added; finds DOI for a given data product and release * zipsByURI() can now handle R object inputs
Bug fixes: * column class matching fixed for lab tables * empty string in API token value handled
Enhancements: * stackByTable(), stackEddy(), byTileAOP(), and byFileAOP() include issue log in outputs * getIssueLog() is also available as an independent function
Bug fixes: * improved error messaging when no data are found * new error messaging if only metadata files are found
Bug fixes: * readme file handling in stackByTable() was failing on new data products; fixed. * clean up of package dependencies
Enhancements: * more options in stackEddy() variable inputs; see function documentation
Bug fixes: * readr package dependency removed to resolve incompatibility with readr 2.0.0 * rare Windows bug for special characters fixed * footRaster() orientation correction in 2.1.0 did not work; corrected again * stackEddy() time stamp handling improved for profile sensors
Enhancements: * download by release in zipsByProduct() and loadByProduct() * per sample files, found in microbe community composition and field spectra, are stacked in stackByTable() * stackFromStore() updated to work with neonstore v0.4.3
Bug fixes: * stackEddy() handles scenarios with only one instance of a given variable * footRaster() coordinate conversion for Alaska and Hawaii fixed * footRaster() orientation corrected * leading digit in 2D wind data tables is translated, with an alert, in loadByProduct() * warning message in stackByTable() when file paths are longer than 260 characters in Windows
Bug fixes: * Release tag assignment in stackByTable() resolved in portal data downloads * Identification of most recent sensor positions files and lab files in stackByTable() resolved * Improved handling of failed downloads in download functions * Empty filler records for single days with no data removed in stackEddy()
Major version update, corresponding to the first Release of static, DOI-citable NEON data. Older versions of neonUtilities may not work correctly with Released data. For more information, see Releases web page.
Major changes: * stackByTable() and stackEddy() updated to work with new zip folder structure * zipsByProduct() updated to use packages API endpoint instead of pre-packaged zip files
Enhancements: * stackByTable() adds release tag to stacked data when possible * getDatatable() is deprecated and functionality is moved to loadByProduct(). Note download by table is not recommended for new users; familiarity with the data is a prerequisite.
Bug fixes: * stackFromStore() now includes the full range of input options for sensor and SAE data * if empty files are downloaded, stackByTable() skips them instead of failing. Note this is rare, the result of error in NEON publication systems. * coordinate conversion for BLAN locations in byTileAOP() updated to use latest versions of spatial packages
Bug fixes: * fix bug in handling of .gz files in stackEddy() * fix pubdate filtering in stackFromStore() * regularize fail behavior when API is unavailable
Bug fixes: * handles .gz files in stackEddy() * footRaster() update to work with latest rgdal
Enhancements: * new stackFromStore() function to stack files from a local archive * aquatic site requests can download meteorological data from nearby terrestrial sites * stackEddy() accepts a vector of filepaths to .h5 files
Bug fixes: * removed gdata package dependency
Enhancements: * input avg to zipsByProduct() changed to timeIndex for clarity
Bug fixes: * unzipped file deletion no longer crashes on large numbers of files
Enhancements: * compatible with new format of sensor positions files * zipsByURI() works with fastq files * lab-specific tables are stacked as well as site-specific tables
Bug fixes: * loadByProduct() automatic deletion of temporary files restored * when API rate limit is reached, zipsByProduct() pauses to reset
Bug fixes: * byFileAOP() and byTileAOP() escape from infinite loops if availability is in error * water quality (DP1.20288.001) error handling fixed * field spectra (DP1.30012.001) handling enabled - AOP functions error cleanly, OS/IS functions proceed * stackEddy() works on expanded package data again * byTileAOP() prevents conversion of coordinates to unreadable formats * byTileAOP() correctly handles sites that cross UTM zones * byFileAOP() and byTileAOP() handle download for sites that are included in other sites’ flight boxes * loadByProduct() handles missing sensor_positions and readme files * encoding specified in readTableNEON()
Enhancements: * download functions include option to use NEON API token * footRaster() scales raster of flux footprint to geographical coordinates * zipsByProduct() regenerates URLs if they expire
Bug fixes: * readme file retention now works with avg input in zipsByProduct() * fields added by stackByTable() now appear in variables file * regularized package citation
Enhancements: * variables, validation, readme, and sensor_positions file names now include data product number
Enhancements: * stackByTable() optionally uses parallel processing * stackByTable() preserves readme and sensor_positions files * stackByTable() appends a new column containing the publication time stamp * stackByTable() no longer requires folder input * loadByProduct() detects data types (numeric, character, date) based on variables file
Bug fixes: * moved rhdf5 from Imports to Suggests for smoother installation * fixed expected encoding in stackByTable() to UTF-8
Enhancements: * added stackEddy() to extract and merge data from flux data HDF5 files (DP4.00200.001)
Bug fixes: * working progress bar in byFileAOP() and byTileAOP()
Enhancements: * initial version of zipsByURI() to download via URLs within data
Bug fixes: * fixed bug in stackByTable() that deleted existing files when using savepath argument * fixed bug in getDatatable() that created incorrect urls
Enhancements: * enabled date and site subsetting in zipsByProduct()
Bug fixes: * fixed bug in stackByTable() that merged tables with overlapping names * fixed bug in zipsByProduct() avg= option that only worked for a subset of data products * fixed bug in stackByTable() that failed unzipping if savepath=filepath
Enhancements: * added getDatatable() * added option to stackByTable() to load files into the R environment * added progress bar to zipsByProduct() * added loadByProduct() to download and load files in one step
byTileAOP() to download only AOP tiles corresponding to certain coordinatesstackByTable()zipsByProduct() option to download only one averaging interval (e.g. only 30-minute files) for speedier download and stackingzipsByProduct() input option to specify file path to save todata.table fread() and rbind() functions, respectively. This speeds up the stacking process.byFileAOP() for downloading large amounts of AOP data using the API; no longer fails if download duration exceeds 24 hours.stackByTable() is used for remote sensing (AOP), eddy covariance, or digital hemipsheric photos.stackByTable()This has two main bug fixes: * Stops stacking or overwriting multiple copies of a lab file that is the same from one monthly zip file to the next. The code now reads from table_types.rda to figure out whether a table is of type site-date, lab-all, lab-current, site-all. This file needs to be updated every time that a new or revised data product is available. * Can properly read new and reprocessed OS data products that use the new file naming convention.