Reading Image Services

ArcGIS Online and Enterprise web services can easily be read into R usingarcgislayers. Supported service types include:

Metadata for all of the above service types can be accessed using arc_open(). Feature data can be read in using arc_select() for FeatureLayer, Table, and ImageServer.

This tutorial will teach you the basics of reading data from hosted image services into R as {terra} SpatRaster objects usingarcgislayers. The source for an image service is published raster or imagery data. To learn more about image services, see the Image services documentation.

Objective

The objective of this tutorial is to teach you how to:

find a image service url from ArcGIS Online
read in the data from the image service
filter the image service by a bounding box
use terra for viewing and writing

Obtaining an image service url

For this example, you will read in multispectral Landsat imagery of the Ouarkziz Crater from ArcGIS Online.

You will use the functions arc_open() and arc_raster() to read image data from ArcGIS Online into R. However, these functions require the url of the hosted image service. To find this, navigate to the item in ArcGIS Online.

When you scroll down, on the right hand side, you will see a button to view the service itself.

view url imagery

Clicking this will bring you to the Image Service, where you can more closely investigate the metadata and supported operations for this service. Navigate to your browser’s search bar and copy the url.

https://landsat2.arcgis.com/arcgis/rest/services/Landsat/MS/ImageServer

Opening an Image Service

First, load the arcgis R package. If you do not have arcgis installed, install it with pak::pak("r-arcgis/arcgis") or install.packages("arcgis").

pak is an R package that makes it faster and easier to install R packages. If you do not have it installed, run install.packages("pak") first.

library(arcgis)

Use the below code to store the image service url in an object called url.

url <- "https://landsat2.arcgis.com/arcgis/rest/services/Landsat/MS/ImageServer"

Then pass this variable to arc_open() and save it to imgsrv (image service).

imgsrv <- arc_open(url)
imgsrv

#> <ImageServer <11 bands, 26 fields>>
#> Name: Landsat/MS
#> Description: Multispectral Landsat image service covering the landmass of the World. This serv
#> Extent: -20037507.07 20037507.84 -9694091.07 9691188.93 (xmin, xmax, ymin, ymax)
#> Resolution: 30 x 30
#> CRS: 3857
#> Capabilities: Catalog,Image,Metadata

arc_open() will create a ImageServer object. Under the hood, this is really just a list containing the image service’s metadata.

The ImageServer object is obtained by adding ?f=json to the image server url and processing the json. All of the metadata is stored in the ImageServer object. You can see this by running unclass(imgsrv). Be warned! It gets messy.

With this ImageServer object, you can read data from the service into R!

Reading from a Image Service

Once you have a ImageServer object, you can access the image data using the arc_raster() function. Pass the coordinates for a bounding box into the function using the xmin, ymin, xmax, and ymax arguments. Store the results of arc_raster() in the object crater.

Avoid reading in more data than you need! When extracting data from an image service, it is best practice to include a bounding box to limit the extraction to just the area that you need. Make sure to provide the bounding box coordinates in the Coordinate Reference System (CRS) of the image service or use the bbox_crs argument to specify another CRS for these coordinates.

crater <- arc_raster(
  imgsrv,
  xmin = -846028,
  ymin = 3373101,
  xmax = -833783,
  ymax = 3380738
)
crater

#> class       : SpatRaster 
#> dimensions  : 400, 400, 11  (nrow, ncol, nlyr)
#> resolution  : 30.6125, 30.6125  (x, y)
#> extent      : -846028, -833783, 3370797, 3383042  (xmin, xmax, ymin, ymax)
#> coord. ref. : WGS 84 / Pseudo-Mercator (EPSG:3857) 
#> source      : file13ff535d7a218.tiff 
#> names       : Coast~rosol, Blue, Green, Red, NearInfrared, Short~red_1, ...

The result is a SpatRaster object that you can now work with using terra and any other R packages.

Using `terra`

From here, you can pursue your own raster and imagery workflows using terra. For some simple examples, consider plotting the image:

terra::plotRGB(crater, stretch = "lin")

or saving the image locally:

terra::writeRaster(crater, "ouarkziz-crater-RGB.tif", overwrite = TRUE)