Introduction to elevatr

Key information about version 0.99.0 and upcoming versions of elevatr

Several major changes have been made to elevatr in response to the retirement of legacy spatial packages (see https://r-spatial.org/r/2023/05/15/evolution4.html for details). Version 0.99.0 has switched to using sf and terra for all data handling; however, in this version a raster RasterLayer is still returned from get_elev_raster(). Additional changes are planned for version 1+, most notably the return for get_elev_raster() will be a terra SpatRaster. Please plan accordingly for your analyses and/or packages account for this change.

Introduction to elevatr

Elevation data is used for a wide array of applications, including, for example, visualization, hydrology, and ecological modelling. Gaining access to these data in R has not had a single interface, is made available through functions across many packages, or requires local access to the data. This is no longer required as a variety of APIs now exist that provide programmatic access to elevation data. The elevatr package was written to standarize access to elevation data from web APIs. This introductory vignette provides details on how to use elevatr to access elevation data and provides a bit of detail on the source data it accesses.

As of version 0.4.2, there are several endpoints that elevatr accesses. For point elevation data it uses USGS Elevation Point Query Service (United States only) as well as the Amazon Web Services (AWS) Terrain Tiles from which point elevations are extracted. Raster elevation data (i.e., Digital Elevation Models or DEMs) are available from the AWS Terrain Tiles or from the OpenTopography Global DEM API (https://portal.opentopography.org/apidocs/#/Public/getGlobalDem) . Currently, elevatr supports the SRTMGL3, SRTMGL1, AW3D30, and SRTM15Plus datasets.

Get Point Elevation Data

Point elevation is accessed from get_elev_point(). This function takes either a data.frame with x (longitude) and y (latitude) locations as the first two columns or a sf, as input and then fetches the reported elevation for that location. As mentioned there is one service that provides this information. Details are provided below.

USGS Elevation Point Query Service

The USGS Elevation Point Query Service is accessible from elevatr. It is only available for the United States (including Alaska and Hawaii). Points that fall within the United States but are not on land return a value of zero. Points outside the United States boundaries return a value of -1000000.

Using get_elev_point() to Access The USGS Elevation Point Query Service

Usage of get_elev_point() requires an input SpatialPoints, SpatialPointsDataFrame, or a two-column data frame with column one containing the x (e.g. longitude) coordinates and the second column containing the y coordinates (e.g. latitude). The source data are global and also include estimates of depth for oceans.

Example usage of each is included below. For these examples, we can create a dataset to use.

# Create an example data.frame
set.seed(65.7)
examp_df <- data.frame(x = runif(3, min = -73, max = -72.5), y = runif(3, min = 42,
    max = 43))
crs_dd <- 4326

# Create and example data.frame with additional columns
cats <- data.frame(category = c("H", "M", "L"))

examp_df2 <- data.frame(examp_df, cats)

# Create an example
examp_sf <- sf::st_as_sf(examp_df2, coords = c("x", "y"), crs = crs_dd)

If a data frame is used it may have additional columns beyond the first two, which must contain the coordinates. The additional columns, along with the returned elevation, will be part of the output POINT or MULTIPOINT sf object. Similarly, an elevation and units column is added to the data frame.

The USGS Elevation Point Query Service returns a single point at a time. The implementation in get_elev_point() will loop through each point, thus can be slow for large number of requests.

Accessing data from this service is done by setting the src to "epqs". No API key is required and there are no rate limits.

df_elev_epqs <- get_elev_point(examp_df, prj = crs_dd, src = "epqs")
df_elev_epqs
df2_elev_epqs <- get_elev_point(examp_df2, prj = crs_dd, src = "epqs")
df2_elev_epqs
sf_elev_epqs <- get_elev_point(examp_sf, src = "epqs")
sf_elev_epqs

Point elevation from Amazon Web Service Terrain Tiles

Since version 0.2.0, elevatr has also provided access to point elevations from the AWS Terrain Tiles. This is not a separate service from the raster DEM service (see below). It is provided as a convenience so users don’t need to download the raster DEMs from AWS and perform their own point data extraction. The added benefit is that points outside of the United States may be used with the AWS source. To access the the point elevations using “aws”:

df_elev_aws <- get_elev_point(examp_df, prj = crs_dd, src = "aws")

An important thing to note, that the elevations will differ, and the prime reason is the resolution of the AWS tiles at the specified zoom. The default zoom of 5 (i.e., z=5) is rather coarse and that is reflected in the elevations.

df_elev_aws$elevation
df_elev_epqs$elevation

A larger zoom results in a smaller pixel size and the two sources converge.

df_elev_aws_z12 <- get_elev_point(examp_df, prj = crs_dd, src = "aws", z = 12)
df_elev_aws_z12$elevation
df_elev_epqs$elevation

Determining the correct zoom is a function of the needs of the user and represents a trade off between higher accuracy/longer downloads.

Lastly, an example using locations outside of the United States.

mt_everest <- data.frame(x = 86.925, y = 27.9881)
everest_aws_elev <- get_elev_point(mt_everest, prj = crs_dd, z = 14, src = "aws")
everest_aws_elev

Get Raster Elevation Data

While point elevations are useful, they will not provide the information required for most elevation based analysis such as hydrologic modeling, viewsheds, etc. To do that requires a raster digital elevation model (DEM). There are several sources for digital elevation models such as the Shuttle Radar Topography Mission (SRTM), the USGS National Elevation Dataset (NED), Global DEM (GDEM), and others. Each of these DEMs has pros and cons for their use. Prior to its closure in January of 2018, Mapzen combined several of these sources to create a synthesis elevation product that utilizes the best available elevation data for a given region at given zoom level. Additionally, the elevation data are enhanced with the inclusion of bathymetry in oceans from ETOPO1. Although closed, these data compiled by Mapzen are made available through two separate APIs: the Nextzen Terrain Tile Service and the Terrain Tiles on Amazon Web Services. Only the Amazon tiles are currently accessible via elevatr.

The input for get_elev_raster() is a data.frame with x (longitude) and y (latitude) locations as the first two columns, any sp object, any sf object, any terra object, or any raster object and it returns a RasterLayer of the tiles that overlap the bounding box of the input. If multiple tiles are retrieved, the resultant output is a merged RasterLayer. Details for each service and their usage via get_elev_raster() are provided below.

Using get_elev_raster() to access the Terrain Tiles on AWS.

As mentioned a data frame with x and y columns, a sp object, or a raster object needs be the input and the src needs to be set to “mapzen” (this is the default).

There is no difference in using the sp and raster input data types. The data frame requires a prj. We show examples using a SpatialPolygonsDataFrame and a data frame. The zoom level (z) defaults to 9 (a trade off between resolution and time for download), but different zoom levels are often desired. For example:

# sf POLYGON example
data(lake)
elevation <- get_elev_raster(lake, z = 9)
plot(elevation)
plot(st_geometry(lake), add = TRUE, col = "blue")

# data.frame example
elevation_df <- get_elev_raster(examp_df, prj = crs_dd, z = 5)
plot(elevation_df)
plot(examp_sf, add = TRUE, col = "black", pch = 19, max.plot = 1)

The zoom level determines the resolution of the output raster. More details on resolution and zoom level is still available in the Mapzen Documentation on ground resolution.

In addition the the required arguments (locations, z, and prj for data frames), several additional arguments may be passed to get_elev_raster(). First, the expand argument is provided to expand the size of the bounding box by a given value in map units. This is useful when bounding box coordinates are near the edge of an xyz tile. For example:

# Bounding box on edge
elev_edge <- get_elev_raster(lake, z = 10)
plot(elev_edge)
plot(st_geometry(lake), add = TRUE, col = "blue")

# Use expand to grab additional tiles
elev_expand <- get_elev_raster(lake, z = 10, expand = 15000)
plot(elev_expand)
plot(st_geometry(lake), add = TRUE, col = "blue")

Second, the clip argument provides some control over the extent and shape of the elevation raster that is returned. The default value returns the entire tile for the “aws” src. The default value for the OpenTopography is also tile, but as these datasets are not served by tile, the “tile” and “bbox” clip return the same elevation raster. The “locations” clip option will clip the elevation raster to the locations themselves. If the input locations are points, this is no different that “bbox”, however if the input locations are a polygon, the elevation raster will be clipped to the boudary of that polygon. For example:

lake_buffer <- st_buffer(lake, 1000)

lake_buffer_elev <- get_elev_raster(lake_buffer, z = 9, clip = "locations")
plot(lake_buffer_elev)
plot(st_geometry(lake), add = TRUE, col = "blue")
plot(st_geometry(lake_buffer), add = TRUE)

Lastly, ... provides the ability to pass additional arguments to httr::GET which is used to access the API endpoints. While any httr::GET arguments may be used, this will most likely be used to pass on configuration arguments such as httr::timeout() or httr::verbose() via a named argument, config to httr::GET. The httr::timeout() can be used to increase the timeout if downloads are timing out. For instance:

library(httr)
# Increase timeout:
get_elev_raster(lake, z = 5, config = timeout(100))

Lastly, multiple configurations may be passed. Below is an example combining httr::timeout() with httr::verbose().

library(httr)
# Increase timeout:
get_elev_raster(lake, z = 5, config = c(verbose(), timeout(5)))

Access OpenTopography API with get_elev_raster()

As of version 0.3.1, the OpenTopography API (https://portal.opentopography.org/apidocs/#/Public/getGlobalDem) has been available as another source of data from get_elev_raster(). To access this data you need to specify which of the available global DEMs you would like to access. The currently available options are: gl3”, “gl1”, “alos”, “srtm15plus”.

Starting in January of 2022, all OpenTopography API requests will require an API key. Version 0.4.3 and greater of elevatr supports this.

To create a key, visit https://portal.opentopography.org/myopentopo. If you do not already have an account, you can create one here: https://portal.opentopography.org/newUser. Once you have your account and are logged into your myOpenTopo Workbench, you can create a new key by scrolling to My Account and selecting “myOpenTopoAuthorization and API Key”. Once there scroll down to the API Key section to create your key.

Within elevatr the API key is expected to be stored as an R environment variable. The easiest way to set this is to use the set_opentopo_key() function. It has a single argument for your API key. Once this is set, restart R and elevatr will use this key for subsequent OpenTopography API requests.

Below is an example for grabbing the OpenTopography SRTM data.

lake_srtmgl1 <- get_elev_raster(lake, src = "gl1", clip = "bbox", expand = 1000)
plot(lake_srtmgl1)
plot(st_geometry(lake), add = TRUE, col = "blue")