Exploring Landsat 8 Data with R
B. Griffith
Exploring Multispectral Imaging Data
1 Introduction
The Landsat 8 satellite was launched in 2013 to collect images of Earth’s surface. It orbits the Earth at an altitude of 438 miles and completes one orbit in 99 minutes. Landsat 8 imaging data supports scientific understanding of Earth’s carbon cycle, water cycle, ecosystems, and various surface changes.
2 Multispectral Data
2.1 Landsat 8 Payload Instrumentation.
The Landsat 8 instrument payload is referred to as the observatory. It includes the Operational Land Imager (OLI) and the Thermal Infrared Sensor (TIRS). Landsat 8 continuously collects data for wavelengths across the spectrum, visible and invisible. The spectral reflections from Earth’s surface allow Landsat 8 to capture phenomena such as vegetation health, water currents, fire damage, and soil moisture content.
2.2 Landsat 8 Raster Data.
The U.S. Geological Survey (USGS) distributes Landsat 8 data. The USGS is a science bureau within the U.S. Dept. of the Interior. Upon request, the USGS distributes a zip folder for a requested area. The zip folder contains a separate raster file (\(tif\)) for each band of the spectrum for an observation area (scene). Short descriptions for spectral bands are provided below. Raster files are typically large. It may be helpful to plot only specific areas or crop files to make them smaller and easier to use.
| Bands | Description | Spatial Resolution | Observe |
|---|---|---|---|
| Band 1 | Deep Blue and Violet | 30 m. | Coastal zone observations; dust; smoke |
| Band 2 | Visible Blue | 30 m. | Distinguishes soil from vegetation; deciduous vs coniferous |
| Band 3 | Visible Green | 30 m. | Emphasizes peak vegetation; plant vigor |
| Band 4 | Visible Red | 30 m. | Emphasizes vegetation on slopes |
| Band 5 | Near-Infrared | 30 m. | Plant health; water; landforms |
| Band 6 | SWIR 1 Shortwave Infrared | 30 m. | Drought stress; burnt areas; active fires; geology |
| Band 7 | SWIR 2 | 30 m. | Drought stress; fire-affected areas; active fires; geology |
| Band 8 | Panchromatic | 15 m. | Sharpens multispectral images |
| Band 9 | Cirrus | 30 m. | Detects cirrus clouds |
| Band 10 | Long wavelength infrared 1 | 100 m. | Ground temp; thermal differences in water currents |
| Band 11 | Long wavelength infrared 2 | 100 m. | Same as band 10 |
2.3 Viewing Files.
The Landsat 8 data folder includes information files (\(xlm\) and \(txt\)) which identify the raster file names for each band, spacecraft ID, upper left and lower right coordinates of the scene, acquisition date, and much more. The Landsat 8 data can be explored with R libraries: raster, rgdal, and rgeos. The separate raster files can be loaded into a raster brick using the \(brick()\) function and plotted in a single set as shown below (bands 1-7).
The images show an area near Nederland, Colorado, in UTM Zone 13. The images from multiple bands can be combined into a single view using \(plotRGB()\). The bands making up the human visual spectrum (2 ,3 , and 4) are combined into one plot below. An area affected by a forest fire can be seen near the center of the image. A body of water can be seen in the lower left quadrant.
Different band combinations can help to clarify a burn area. A false color composite using bands 6, 5, and 3 is shown below. Notice the improved clarity of the burn border and the high contrast between land and water.
The ESRI.com website includes recommendations for viewing vegetation health by combining bands 5, 4, and 3, with highest vigor showing in red.
Another vibrant false color composite for visualizing forests and fire burns can be produced by combining bands 7, 5, and 2.
Terrain images produced by Landsat data can be affected by cloud cover. Notice the effects of cloud cover and shadow in the following scene captured on a different day.
Finally, we examine the Colorado Cold Springs Fire burn area using the near infrared (NIR: band 5) and the shortwave infrared (SWIR: band 7) bands combined. This will provide a surface-reflectance-based value used to identify burned areas and provide a measure of burn severity. We will calculate the Normalized Burn Ratio (NBR) as: (NIR - SWIR)/(NIR + SWIR) or (Band 5 – Band 7) / (Band 5 + Band 7) and plot the resulting view.
3 Conclusions
Landsat has provided years of imaging data for broad-ranging areas of study including agriculture, oceanography, glaciology, forestry, and urban development. The data is available and free of charge from the USGS. While R programming aids exploration, the USGS EarthExplorer website provides data viewing capabilities without the need for programming to encourage audiences of all backgrounds.
Data Source: From EarthDataScience.org: Earth Data Analytics, Univ. of Colorado, Boulder. Other Resources: Definitions and illustration of Landsat 8 from USGS.gov https://earthexplorer.usgs.gov/ 03/2021