This morning, Ned Horning provided a tutorial on downloading Landsat imagery to Cape Cod Baywatch. Cape Cod Baywatch's specific interest is in acquiring thermal imagery of the Cape Cod Bay, in order to assess the environmental impacts of the thermal plume caused by the Plymouth nuclear power plant; but there are many other applications for Landsat imagery in the Public Lab community, too. The tutorial, linked to below, covers:
- Navigating to the USGS satellite imagery archive website: http://glovis.usgs.gov/
- Locating a preview of the Landsat imagery for a specific historical date, and location
- Registering on the site, to enable downloading the high resolution image layers (visible band, thermal, etc)
- Selecting and downloading the imagery
- Opening up the images in a QGIS (open source GIS software)
- Changing display settings (contrast, false color) in order to highlight features of interest (in this case, a thermal plume in the Bay water)
- Extracting information from the image metadata in order to correlate thermal image brightness levels with temperature
Here it is:
Details on temperature calibration
As a follow-up, Ned sent along the following information:
As far as calculating brightness temperature you can use these two equations:
Top of the atmosphere radiance: TOA Radiance = DN * RADIANCE_MULT_BAND_10 + RADIANCE_ADD_BAND_10
DN is the pixel values in the image.
Here is the formula that I used in QGIS Raster Calculator: "LC80120312014173LGN00_B10@1" * 0.0003342 + 0.1
You would replace "LC80120312014173LGN00_B10@1" with your image.
To calculate brightness temperature use this equation:
T = K2_CONSTANT_BAND_10 / (ln((K1_CONSTANT_BAND_10 / TOA Radiance) + 1))
If you want to combine both of these so you only have to run it once the equation is: T = K2_CONSTANT_BAND_10 / (ln((K1_CONSTANT_BAND_10 / (DN * RADIANCE_MULT_BAND_10 + RADIANCE_ADD_BAND_10)) + 1))
Here is the formula that should work in the QGIS Raster Calculator but it doesn't since it appears that the natural logarithm, "ln", isn't supported yet in this new version of the raster calculator. The user guide mentions that it will be supported in a future release.
1321.08 / ln(774.89 / ("LC80120312014173LGN00_B10@1" * 0.0003342 + 0.1) + 1)
Next steps ...
Next, we're hoping to dig more into pulling out meaningful temperature values from the thermal images, and talking with Ned more about developing scripts (in R, say) that go some way towards automating some of the analysis we'd like to do on the images.
Thanks to Ned for a great, easy-to-understand tutorial! And for the great discussion and ideas generated by Karen, Pine, Alex, and Adam.