Band Combinations for Landsat 8

Landsat 8 has been online for a couple of months now, and the images look incredible. While all of the bands from previous Landsat missions are still incorporated, there are a couple of new ones, such as the coastal blue band water penetration/aerosol detection and the cirrus cloud band for cloud masking and other applications. Here’s a rundown of some common band combinations applied to Landsat 8, displayed as a red, green, blue (RGB):

Natural Color 4 3 2
False Color (urban) 7 6 4
Color Infrared (vegetation) 5 4 3
Agriculture 6 5 2
Atmospheric Penetration 7 6 5
Healthy Vegetation 5 6 2
Land/Water 5 6 4
Natural With Atmospheric Removal 7 5 3
Shortwave Infrared 7 5 4
Vegetation Analysis 6 5 4

Here’s how the new bands from Landsat 8 line up with Landsat 7:

Landsat 7

Landsat 8

Band Name Bandwidth (µm) Resolution (m) Band Name Bandwidth (µm) Resolution (m)
Band 1 Coastal

0.43 – 0.45

30

Band 1 Blue

0.45 – 0.52

30

Band 2 Blue

0.45 – 0.51

30

Band 2 Green

0.52 – 0.60

30

Band 3 Green

0.53 – 0.59

30

Band 3 Red

0.63 – 0.69

30

Band 4 Red

0.64 – 0.67

30

Band 4 NIR

0.77 – 0.90

30

Band 5 NIR

0.85 – 0.88

30

Band 5 SWIR 1

1.55 – 1.75

30

Band 6 SWIR 1

1.57 – 1.65

30

Band 7 SWIR 2

2.09 – 2.35

30

Band 7 SWIR 2

2.11 – 2.29

30

Band 8 Pan

0.52 – 0.90

15

Band 8 Pan

0.50 – 0.68

15

Band 9 Cirrus

1.36 – 1.38

30

Band 6 TIR

10.40 – 12.50

30/60

Band 10 TIRS 1

10.6 – 11.19

100

Band 11 TIRS 2

11.5 – 12.51

100

For the most part, the bands line up with what we’re used to, with some minor tweaking of the spectral ranges. The thermal infrared band from Landsat 7 is now split into two bands for Landsat 8. Whereas before you had one thermal band that was acquired at 60 m resolution (and resampled to 30 m) now you have increased spectral resolution at the cost of spatial resolution. It wouldn’t be remote sensing without tradeoffs, right?

This entry was posted in Imagery, Uncategorized and tagged , , . Bookmark the permalink.

Leave a Reply

17 Comments

  1. franzpc says:

    Do you know if I can make a DEM with Band 8 Pan? How to …
    Thanks

    • kevin_butler says:

      Short answer is no. Landsat would have to have a forward and backward looking band to generate a stereo image for dem purposes. All of its bands are at nadir. Check out Aster (30 m globally) or SRTM (90 m globally).

  2. joabelb says:

    Hi Kevin. First of all congratulations for this excellent post.
    I would like to know if ESRI is planning to release a basemap of Landsat 8 images (time enabled).
    I just saw something like that in this website http://landsatlook.usgs.gov/ . It would be awesome to have this basemap on my ArcGIS Desktop…
    tks

  3. alexanderwandl says:

    Hi Kevin,
    I have a question concerning the calculation of the NDVI. switching the bands 3 to 4 and 4 to 5 in the image analyses extension doesn’t provide reasonable results, any advice?
    Thanks
    Alex

  4. horizonsweb says:

    How can I combine the Landsat 8 bands in ArcGIS 10.1?

    • kevin_butler says:

      If you’re working with one Landsat scene, load all of the bands onto your map. From the Image Analysis Window, select all of them and then click the composite bands function in the processing pane.
      If you have multiple scenes, the easiest way to do this without scripting anything is to create a mosaic dataset of each band and then use composite bands on each of the mosaic datasets.

  5. William says:

    Hello

    May I ask what is the best band combination to display a scene where the land cover includes forest, bare soil, crops and urban areas

    • kevin_butler says:

      If the true color composite doesn’t work for you, I would try something along the lines of 7 6 4. Bands 2 or 3 could also work in the blue band. If the vegetation isn’t sufficiently highlighted, try switching the green band from 6 to 5.

  6. naveenjayanna says:

    How to Layer Stack Landsat 8 bands……

    plz respond soon……

    Thank you

  7. gauravhegde24 says:

    Hello Kevin,

    I needed guidelines on how a particular object/class looks like in various band combinations for landsat images.
    for example crop looks bright red, fallow land looks yelow in fcc..

    I need this for image classification. Hope I can get some reference for all major band combinations.

    • kevin_butler says:

      I hate to break it to you, but it’s really not that simple. There are a ton of factors that go into how features look, and if you want to do a classification, me telling you that vegetation is bright in the NIR is not going to get you where you want to be. My suggestion is load up the imagery basemap from Arcgis Online and use that as a reference. Some stuff may have changed if there is a big difference in when the landsat imagery and the highres imagery was acquired, but most things are consistent. When you see a feature that you’re interested in, start playing with the band combinations and get a feel for how those features change as you change the band combinations. What are you classifying for?

      • gauravhegde24 says:

        Thanks for the suggestions Kevin.
        I am trying to classify a Landsat8 image to detect temporal changes in a mining area.
        I am trying to figure out how to be sure about signature pixels I choose..

        Regards,
        Gaurav

        • kevin_butler says:

          You won’t be sure about your training sites until you actually run the classification. It tends to be an iterative process that involves a lot of tweaking to the training sites until you are satisfied with the output. Start with pixels that seem representative of the feature you want to classify. If they are too “pure” you’ll miss classify some of the mixed pixels. If they’re too “mixed” you’ll miss out on the pixels that only have that feature. It becomes as much of an art as science at this point.

          If you use the MLClassify function (from the image analysis window, not the geoprocessing tool) it will display your results on the fly. I find that this saves time when I’m going through and trying to perfect a classification.