Geospatial Data

Physical World → GIS World

Start by chopping up the real world into different “layers” with similar features:

Option 1 is to represent features as “vectors” (points, lines and polygons):

Option 2 is to represent features as a grid or “raster”:

Projections

How do we squish a round planet onto flat maps?

UTM

A common projection system is Universal Transverse Mercator (UTM). UTM divides the planet up into 60 zones that are 6 degrees wide. The USA falls in zones 10-19.

What You Need to Remember about Projections in R

With rare exceptions (i.e., tmap), R does not reproject on the fly
The sf package can read & write projection info when importing or exporting GIS files
If your analysis uses distance or area in any way, you should use projected coordinates (not lat-long)
Some web map packages (e.g., leaflet) require lat-long
(Re)projecting is easy (with st_transform())
The best way to refer to projections is by EPSG numbers

Attributes

Every feature has some additional, non-spatial properties, typically called attributes.

Attributes are used for:

symbology
selecting features (aka, attribute query)
bringing in info from other tables (aka attribute join)
analysis

Raster Data

Remotely Sensed Data

satellite sensors
airborne or UAV mounted
DEMs from radar

Landsat infrared image (false-color-composite)

Elevation (measured from Shuttle)

Derived or modeled data

interpolated surfaces (e.g., environmental variables, elevation)
Kernel density surface (probability of occurrence)
cost surfaces
classified images

Land cover map (derived from satellite image)

Distance to trails (computed)

Single vs. Multi-Band Rasters

Rasters with multiple layers are common in multi-spectral data (e.g., Landsat). Each band saves the reflectance in a different part of the electromagnetic spectrum.

Whether saved as separate RasterLayer objects, or a single multi-layered RasterStack or RasterBrick, up to three layers can be combined to produce colors when plotting.

Raster Spatial Properties

Raster data structures are always rectangular. The spatial structure is implicit.

Spatial properties include:

origin
projection / CRS
resolution (pixel size)

Pixel Values: Continuous vs Discrete

Cell values can be either continuous or discrete.

Continuous

Discrete / Categorical

Discrete rasters also have a ‘Raster Attribute Table’ (RAT), which stores the name of each category.

Rasters vs. Tiles

Rasters are data you can visualize and manipulate.

Tiles are background images you get from online providers for display purposes. They look nice, but what you see is what you get (no ability to manipulate).

Continuous or Discrete?

How would you represent each of the following surfaces - continuous or discrete?

temperature
land cover
land use
soil type
fire risk

Next: Files and Folders

Spatial Data Analysis with RSociety for Conservation GIS, July 2021