Spatial Data Vector

Remember to download and put into data subdirectory:

LiDAR rasters and plot locations

Load the following into browser window:

Projections picture

library(ggplot2)
library(sf)
library(stars)

We’ll also load ggplot2 again for plotting
For raster data we’ve loaded it using read_stars and plotted it with geom_stars

dtm_harv <- read_stars("data/HARV/HARV_dtmCrop.tif")
ggplot() +
  geom_stars(data = dtm_harv)

plots_harv <- st_read("data/HARV/harv_plots.shp")
ggplot() +
  geom_sf(data = plots_harv)

ggplot() +
  geom_stars(data = dtm_harv) +
    geom_sf(data = plots_harv)

That wasn’t what we expected
We don’t see the raster data and there appears to just be one point and an empty map
Why?

The reason this graph doesn’t work is that the two datasets have different projections
We can see this by going back to the individual plots
The axes on the vector plot latitude and longitude values in degrees, with numbers in the low 40s and low 70s
The axes on the raster plot are much different, with values in the hundreds of thousands
These differences are because the two sets of data have different “coordinate reference systems” or “projections”
Since the earth is round we have to stretch geospatial data to present it on flat maps
There is no one best way to do this so there are different projects, which result in different representations of the world, and different units for locations
Here are examples of a few common ones including two we’ll be working with WGS 84, which is latitude & longitude, and UTM

Map of the United States in four projections. Mercator, U.S. National Atlas Equal Area, UTM Zone 11N, and WGS 84. The maps all appear different.

The “coordinate reference system” or “CRS” indicates how this is done
Coordinate Reference System (crs or projection) is different from raster.
We can use st_crs to look up the CRS for this spatial data

st_crs(dtm_harv)

st_crs(plots_harv)

To work with data having different projections together we can transform the projections to match each other
Do this using the st_transform function
Takes two arguments
The geospatial object to be transformed
The CRS to transform it to
There are a variety of ways to indicate a CRS
Including numeric codes and “well known text” of WKT representations for different coordinate reference systems
E.g., the code 4326 indicates the standard longitude latitude CRS, so we could transform our raster data to lat longs

dtm_lat_long <- st_transform(dtm_harv, 4326)
ggplot() +
  geom_stars(data = dtm_harv)

Often the easiest thing to do when combining geospatial data is to match all objects to one of the existing CRS’s
Do this by running st_crs on the object whose CRS you want to match
So we’ll transform our plots data to have the same CRS as our vector data

plots_harv_utm <- st_transform(plots_harv, st_crs(dtm_harv))

Because these two objects now have the same CRS the plot will look like we’d hoped it would

ggplot() +
  geom_stars(data = dtm_harv) +
  geom_sf(data = plots_harv)

We’ve seen two of the most common CRS’s
Because we’re flattening a sphere no projection is perfect for all circumstances
When choosing a CRS you want to think about what aspects of the world you want to preserve, like distance or area
UTM, which stands for Universal Transverse Mercator, is one of the most commonly used projections in ecological research
It accuractely represents local geospatial information and preserves distance
It is primarily designed to work within different zones and so isn’t generally used at scales larger than a state
Lat-longs are a common way of collecting data, but don’t preserve any key aspects of the data
The Azimuthal Equal Area projection maintains area, so if the amount of area being worked with is important it’s a good projection
So, for most of you UTM within your research zone will be the right way to go
If you work at larger scales think about what it is most important to preserve and look for a transformation that does that

To represent geospatial information from the surface of the sphere-like earth we have to stretch it to make it flat
We do this using projections and that are represented as “coordinate reference systems” or “CRS”
The st_transform function can transform data from one CRS to another
This is often important for working with multiple geospatial objects
Thought in some cases geospatial tools will quietly handle reprojection for us
UTM with an appropriate local zone is the most common CRS used in ecological research

Do Task 3 of Canopy Height from Space.

Notes