R best environnement for spatial analysis Geography is an applied field where we can use our skills in th real world.

Evidence based on decision making.

Robin: transport geographer

Jakub : environnmental data, raster data

You can think in code and test ideas.

Use of reprex: learn code takes times, be able to ask question is important. If something does not work it is important to show what I tried. => Reproducable exemple

rstudioapi::openProject() : open or create a project

GIS ecosystem of R

Wild range of fields where R spatial can be use

Large range of applications - Exploratory data analysis (EDA) - output in different format - data processing - data visualisation

R is a (geo)power tool

R’s ecosystem makes it easy to communicate your results.

Rpubs to share R document (dépot géré par Rstudio, un bouton intégré dans Rstudio)

Building your own packages and distributing code

Examples of packages: - data package : {stats19} - methods package : {noise-phenomena}, {sabre}

R spatial : everything related to spatial data in R

GEOS works well for projected data S2 works for geographical data

Data models

Vector data model

sf implements the Simple Features OGC standard sp maintenance will stop in 2024.

most functions start with an st_ prefix.

terra packages has its own vector data classes but Jakub recommends using the regular geospatial data formats.

Sf reading : - from file (through GDAL) - from CSV with st_as_sf(df, coords = c("X", "Y"))

Raster data model

raster is successed by terra

If you have to change more than 20% of a code base, you better have to start from 0 with a new package.

terra is more performant than raster, most functions are written in C++.

stars contains classes and methods for representing spatiotemporal data (raster and vector data)

rast() reads raster data

Mapping ecosystem

static maps
interactive maps
both (tmap)
specific maps
- {cartogram}
- {geofacet} : divide spatial data into plots (while keeping the overall shape)
- {edgebundle}

Coordinate reference systems

geographic coordinates (GCRS): ellipsoid and datum
projected coordinates (PCRS): GCRS +

How to describe: - CRS codes (like EPSG) - WKT2

The proj4string is discouraged, especially for storing data.

WKT2 is better than proj4string because of accuracy and time robusteness (dur to plaques tectonics)

functions: - sf::st_crs() access to informations : st_crs(world)$epsg - terra::crs()

GIS and mapping - A lecture on the GIS and mapping ecosystem of R

Jakub Nowosad