Geo-spatial and Statistical Packages

A comparison of some key R and Python packages.

Purpose	Package	Key Functions
Vector data handling (points, lines, polygons)	geopandas (Python) sf (R)	`read_file()`, `sjoin()`, `to_crs()`, `dissolve()` `st_read()`, `st_join()`, `st_transform()`, `st_union()`, `st_as_sf()`
Raster data handling (reading/writing gridded data)	rasterio (Python) terra (R)	`open()`, `read()`, `mask()`, `warp.reproject()` `rast()`, `vect()`, `crop()`, `mask()`, `extract()`
Multi-dimensional arrays (NetCDF, climate data)	xarray (Python) terra (R)	`open_dataset()`, `sel()`, `mean()` `rast()`, `app()`, `subset()`
Geospatial xarray operations	rioxarray (Python)	`open_rasterio()`, `rio.reproject()`, `rio.clip()`
Vector-raster operations	xvec (Python) rasterstats (Python) terra (R)	`zonal_stats()`, `extract_points()` `zonal_stats()` `extract()`
Fast raster extraction to polygons (area-weighted)	exactextract (Python) exactextractr (R)	`exact_extract()` `exact_extract()`
Static visualization and plotting	matplotlib (Python) ggplot2 (R)	`pyplot.plot()`, `pyplot.imshow()` `ggplot()`, `geom_sf()`, `geom_raster()`
Interactive web maps	folium (Python) tmap (R) leaflet (R)	`Map()`, `Choropleth()` `tm_shape()`, `tm_polygons()`, `tm_layout()` `leaflet()`, `addTiles()`, `addPolygons()`
Statistical modeling and econometrics	statsmodels (Python) linearmodels (Python) fixest (R) pyfixest (Python)	`OLS()`, `summary()` `PanelOLS()` `feols()`, `etable()` `feols()`, `etable()`
Google Earth Engine access	ee (Python)* rgeedim (R)* rgee (R)*	`ee.Initialize()`, `ee.ImageCollection()`, `ee.Image.reduceRegion()` Download GEE as GeoTIFFs R wrapper for GEE
Animated visualizations	matplotlib.animation (Python)* plotly (Python)* gganimate (R)*	`FuncAnimation()`, `save()` `px.scatter()`, `animation_frame=` Create GIFs from ggplot2
3D visualizations	pyvista (Python)* plotly (Python)* rayshader (R)*	`plot()`, `add_mesh()` `Surface()`, `Scatter3d()` 3D maps and renderings
Geographic utilities	country_converter (Python)* countrycode (R)*	`convert()`, `continent()` Convert country names/codes
Parallel processing	dask (Python)* doparallel (R)*	`compute()`, `delayed()`, `array()` Parallel processing for R

Note

Some of these Packages marked will not be used in this workshop but are useful for advanced workflows.

Important

Important packages for raster data in R.

terra: the best way of loading and processing raster data.
sf: the go-to package for dealing with polygons, lines and spatial points.
exactextractor: the quickest, easiest and most accurate way to calculate zonal statistics. It’s an R wrapper for the exactextract package (written in C++). It’s more accurate than ArcGIS, QGIS, and Python’s rasterstats package in calculating zonal stats.
tmap: quick and easy, and its “view” mode allows you to create interactive maps that are built on leaflet.
ggplot: the best package for building beautiful maps. Allows for more customisation than tmap.

There are also the following packages which we will not be using in this workshop:

rgeedim: provides the easiest way to download data directly from Google Earth Engine as GeoTiffs.
rgee: an R wrapper for Google Earth Engine, so you can run GEE commands in R
gganimate: a package that allows you to turn your ggplot2 graphs into gifs.
rayshader: create animated 3D graphs and maps in R.
countrycode: a useful package to automatically generate ISO codes from country names (and vice versa), assign continents to countries.
doparallel: my go-to package for parallel processing.