Introduction to MazamaSpatialUtils

Mazama Science

Dec 01, 2020

Background

The MazamaSpatialUtils package was created by MazamaScience to regularize our work with spatial data. The sp, rgdal and maptools packages have made it much easier to work with spatial data found in shapefiles. Many sources of shapefile data are available and can be used to make beautiful maps in R. Unfortunately, the data attached to these datasets, even when fairly complete, often lacks standardized identifiers such as the ISO 3166-1 alpha-2 encodings for countries. Maddeningly, even when these ISO codes are used, the dataframe column in which they are stored does not have a standardized name. It may be called ISO or ISO2 or alpha or COUNTRY or any of a dozen other names we have seen.

While many mapping packages provide ‘natural’ naming of countries, those who wish to develop operational, GIS-like systems need something that is both standardized and language-independent. The ISO 3166-1 alpha-2 encodings have emerged as the de facto standard for this sort of work. In similar fashion, ISO 3166-2 alpha-2 encodings are available for the next administrative level down – state/province/oblast, etc. For time zones, the de facto standard is the set of Olson time zones used in all UNIX systems.

The main goal of this package is to create an internally standardized set of spatial data that we can use in various projects. Along with three built-in datasets, this package provides convert~() functions for other spatial datasets that we currently use. These convert functions all follow the same recipe:

Other datasets can be added following the same procedure.

The ‘package internal standards’ are very simple.

  1. Every spatial dataset must contain the following columns:
  1. Spatial datasets with time zone data must contain the following column:
  1. Spatial datasets at scales smaller than the nation-state should contain the following column:

If other columns contain these data, those columns must be renamed or duplicated with the internally standardized name. This simple level of consistency makes it possible to generate maps for any data that is ISO encoded. It also makes it possible to create functions that return the country, state or time zone associated with a set of locations.

Functionality

The core functionality for which this package was developed is determining spatial information associated with a set of locations.

Current functionality includes the following:

A generic getSpatialData(longitude, latitude, ...) returns a dataframe whose rows are associated with specified locations. This function can be used with newly converted SpatialPolygonsDataFrames.

For those working with geo-located data, the ability to enhance location metadata with this information can be extremely helpful.

Standard Datasets and Setup

When using MazamaSpatialUtils, always run setSpatialDataDir(<spatial_data_directory>) first. This sets the directory where spatial data will be installed and loaded from. This can be a directory on a user’s personal computer or perhaps a remotely mounted disk if huge spatial datasets are going to be used.

MazamaSpatialUtils has three built-in datasets:

Version 0.7 of the package is built around the three built-in datasets and several other core datasets that may be installed including:

Install these one at a time with:

setSpatialDataDir('~/Data/Spatial')
installSpatialData("<datasetName>")

Once datasets have been installed, loadSpatialData() can be used load datasets found in the SpatialDataDir that match a particular pattern, e.g:

loadSpatialData('USCensusStates')
loadSpatialData('USCensusCounties')

Additional watershed data from the Watershed Boundary Dataset are quite large and may be installed in the same way:

getCountry() and getCountryCode()

These two functions are used for assigning countries to one or many locations. getCountry() returns English country names and getCountryCode() returns the ISO-3166 two character country code. Both functions can be passed allData = TRUE which returns a dataframe with more information on the countries. You can also specify countryCodes = c(<codes>) to speedup searching by restricting the search to polygons associated within those countries.

These functions use the package-internal SimpleCountries dataset which can be used without loading any additional datasets.

In this example we’ll find the countries underneath a vector of points:

library(MazamaSpatialUtils)
## Loading required package: sp
longitude <- c(-122.3, -73.5, 21.1, 2.5)
latitude <- c(47.5, 40.75, 52.1, 48.5)

# Get countries/codes associated with locations
getCountry(longitude, latitude)
## [1] "United States" "United States" "Poland"        "France"
getCountryCode(longitude, latitude)
## [1] "US" "US" "PL" "FR"
# Review all available data
getCountry(longitude, latitude, allData = TRUE)
##   countryCode   countryName polygonID
## 1          US United States       113
## 2          US United States       113
## 3          PL        Poland       205
## 4          FR        France       167

getState() and getStateCode()

Similar to above, these functions return state names and ISO 3166 codes. They also take the same arguments. Adding the countryCodes argument is more important for getState() and getStateCode() because the NaturalEarthAdm1 dataset is fairly large. Lets use the same latitude and longitude variables as above and assign states to those locations.

These functions require installation of the large NaturalEarthAdm1 dataset which is not distributed with the package.

(The next block of code is not evaluated in the vignette.)

# Load states dataset if you haven't already
loadSpatialData('NaturalEarthAdm1')

# Get country codes associated with locations
countryCodes <- getCountryCode(longitude, latitude)

# Pass the countryCodes as an argument to speed everything up
getState(longitude, latitude, countryCodes = countryCodes)
getStateCode(longitude, latitude, countryCodes = countryCodes)

# This is a very detailed dataset so we'll grab a few important columns
states <- getState(longitude, latitude, allData = TRUE, countryCodes = countryCodes)
states[c('countryCode', 'stateCode', 'stateName')]

getTimezone()

Returns the Olsen time zone where the given points are located. Arguments are the same as the previous functions. allData = TRUE will return other useful information such as the UTC Offset.

These functions use the package-internal SimpleTimezones dataset which can be used without loading any additional datasets.

# Find the time zones the points are in
getTimezone(longitude, latitude)
## [1] "America/Los_Angeles" "America/New_York"    "Europe/Warsaw"      
## [4] "Europe/Paris"
# Get country codes associated with locations
countryCodes <- getCountryCode(longitude, latitude)

# Pass the countryCodes as an argument to potentially speed things up
getTimezone(longitude, latitude, countryCodes = countryCodes)
## [1] "America/Los_Angeles" "America/New_York"    "Europe/Warsaw"      
## [4] "Europe/Paris"
# Review all available data
getTimezone(longitude, latitude, allData = TRUE, countryCodes = countryCodes)
##              timezone UTC_offset UTC_DST_offset countryCode   longitude
## 1 America/Los_Angeles         -8             -7          US -118.242778
## 2    America/New_York         -5             -4          US  -74.006389
## 3       Europe/Warsaw          1              2          PL   21.000000
## 4        Europe/Paris          1              2          FR    2.333333
##   latitude    status notes           polygonID
## 1 34.05083 Canonical       America/Los_Angeles
## 2 40.71167 Canonical          America/New_York
## 3 52.25417 Canonical             Europe/Warsaw
## 4 48.88111 Canonical              Europe/Paris

getUSCounty()

Returns the US County which name pairs of coordinates fall in. The arguments are similar as above except that stateCodes=c() is used instead of countryCodes=c() since this dataset is US specific.

(The next block of code is not evaluated in the vignette.)

# Load counties dataset if you haven't already
loadSpatialData("USCensusCounties")

# New dataset of points only in the US
stateCodes <- getStateCode(longitude,latitude)

# Optionally pass the stateCodes as an argument to speed everything up
getUSCounty(longitude, latitude, stateCodes = stateCodes)
getUSCounty(longitude, latitude, allData = TRUE, stateCodes = stateCodes)

Timezone Map

While identifying the states, countries and time zones associated with a set of locations is important, we can also generate some quick eye candy with these datasets. Let’s color the time zones by the data variable ‘UTC_offset’

library(sp)         # For spatial plotting

# Assign time zones polygons an index based on UTC_offset
colorIndices <- .bincode(SimpleTimezones@data$UTC_offset, breaks = seq(-12.5,12.5,1))

# Color our time zones by UTC_offset
plot(SimpleTimezones, col = rainbow(25)[colorIndices])
title(line = 0, 'Timezone Offsets from UTC')

Working with ISO 3166-1 Encoded Data

On of the main reasons for ensuring that our spatial datasets use ISO encoding is that it makes it easy to generate plots with any datasets that use that encoding. Here is a slightly more involved example using Energy data from the British Petroleum Statistical Review that has been ISO-encoded.

library(sp)         # For spatial plotting

# Read in ISO-encoded oil production and consumption data
prod <- read.csv(url('http://mazamascience.com/OilExport/BP_2016_oil_production_bbl.csv'),
                 skip = 6, stringsAsFactors = FALSE, na.strings = 'na')
cons <- read.csv(url('http://mazamascience.com/OilExport/BP_2016_oil_consumption_bbl.csv'),
                 skip = 6, stringsAsFactors = FALSE, na.strings = 'na')

# Only work with ISO-encoded columns of data
prodCountryCodes <- names(prod)[ stringr::str_length(names(prod)) == 2 ]
consCountryCodes <- names(cons)[ stringr::str_length(names(cons)) == 2 ]

# Use the last row (most recent data)
lastRow <- nrow(prod)
year <- prod$YEAR[lastRow]

# Neither dataframe contains all countries so create four categories based on the
# amount of information we have:  netExporters, netImporters, exportOnly, importOnly
sharedCountryCodes <- intersect(prodCountryCodes,consCountryCodes)
net <- prod[lastRow, sharedCountryCodes] - cons[lastRow, sharedCountryCodes]

# Find codes associated with each category
netExportCodes <- sharedCountryCodes[net > 0]
netImportCodes <- sharedCountryCodes[net <= 0]
exportOnlyCodes <- setdiff(prodCountryCodes,consCountryCodes)
importOnlyCodes <- setdiff(consCountryCodes,prodCountryCodes)

# Create a logical 'mask' associated with each category
netExportMask <- SimpleCountries@data$countryCode %in% netExportCodes
netImportMask <- SimpleCountries@data$countryCode %in% netImportCodes
onlyExportMask <- SimpleCountries@data$countryCode %in% exportOnlyCodes
onlyImportMask <- SimpleCountries@data$countryCode %in% importOnlyCodes

color_export = '#40CC90'
color_import = '#EE5555'
color_missing = 'gray90'

# Base plot (without Antarctica)
notAQ <- SimpleCountries@data$countryCode != 'AQ'
plot(SimpleCountries[notAQ,], col = color_missing)

plot(SimpleCountries[netExportMask,], col = color_export, add = TRUE)
plot(SimpleCountries[onlyExportMask,], col = color_export, add = TRUE)
plot(SimpleCountries[netImportMask,], col = color_import, add = TRUE)
plot(SimpleCountries[onlyImportMask,], col = color_import, add = TRUE)

legend(
  'bottomleft',
  legend = c('Net Exporters','Net Importers'),
  fill = c(color_export,color_import)
)
title(line = 0, paste('World Crude Oil in', year))