What do colours mean in satellite imagery?

False colour image of phytoplankton blooming off the coast of Patagonia. Acquired 2nd Dec 2014. Image Courtesy of NASA/NASA's Earth Observatory

Phytoplankton blooming off the coast of Patagonia on 2nd Dec 2014.
Image Courtesy of NASA/NASA’s Earth Observatory

Satellite images are a kaleidoscope of colours, all vying for attention. It’s important to be clear what the colours are showing, and more importantly, what they may not be showing, to interpret the image correctly. For example, a patch of white on an image might indicate snow or ice, sunglint off the ocean, fog or it could just mean it was cloudy.

On the earth’s surface different colours represent different land types:

  • Vegetation appears as shades of green from pale for grasslands to dark for forests – although some forests will progress from green to orange to brown in autumn.
  • Ocean colour is significantly influenced by phytoplankton, which can produce a range of blue and green colours. A fantastic example of this can be seen in the image at the top of the blog showing phytoplankton blooming off the cost of Patagonia.
  • Snow and ice can appear white, grey, or slightly blue.

As noted in the opening, colours can also mislead with cloud cover being the natural nemesis of optical remote sensing. However, you also have to be careful with effects such as:

  • Smoke: ranges from brown to grey to black.
  • Haze: a pale grey or a dirty white.
  • Dust: can be brown, like bare ground, but also white, red and black.
  • Shadow: Clouds or mountain shadows can look like dark surface features.

There is a good article here from NASA’s Earth Observatory giving more details on the different colours of surface land types. So far, we’ve focussed on natural colour signatures; but man-made structures also appear on imagery. Generally, urban areas tend to be silver or grey in colour; although larger objects also show up in their own right such as the bright red roof of Ferrari World in the middle of the Abu Dhabi Grand Prix Circuit – as discussed in a previous blog.

Composite Google Earth image of the entrance to the Panama Canal: Data courtesy of DigitalGlobe

Composite Google Earth image of the entrance to the Panama Canal: Data courtesy of DigitalGlobe

We tried to repeat the identification of man-made objects for this blog using the coloured roofs of the Biomuseo building, located on the Amador Causeway – at the entrance to the Panama Canal in the Pacific Ocean. Sadly, Landsat 8 pixels are too coarse; and Google Earth has fallen prey to cloud cover preventing visibility, as shown in the image on the right. What you can see though is the buildings in Panama City and the yachts in the marinas and clustered around the four islands (Naos, Perico, Culebra and Flamenco) at the end of the Amador Causeway.

The final thing to remember when considering colours, is the format of the image itself. Some images use true-colours from the red, green and blue wavelengths, which produce colours as if you were looking at the scene directly, so trees are green, sea is blue, etc. However, other images incorporate infrared light to enhance the detection of features not easily distinguished on a true-colour image; this means colours aren’t what you would expect, for example, the ocean may appear red.

Colour is central to use of satellite imagery, but you need to know the properties of the rainbow you are looking at or you may never find the pot of satellite gold.

Completing the PhD publication triple

Some great news this week! Dr Susan Kay’s third paper from her PhD has been accepted for publication by Applied Optics. Entitled “Sun glint estimation in marine satellite images: a comparison of results from calculation and radiative transfer modeling”, it nicely shows the impact of choosing different models for the sea surface elevation and slope when predicting sun glint. In response to the notice of publication Sue said “It’s great to see that last bit of PhD work finished. Now I’d better get writing about marine ecosystem modelling!” which is her current research at Plymouth Marine Laboratory.

I’ve been one of Sue’s PhD supervisors, alongside Dr John Hedley, and it’s wonderful that she’s had three papers published. There is always an expectation that PhD students will produce papers during their studies, on top of writing up their PhD. However peer-reviewed publications aren’t easy to achieve, more and more papers are being produced but scientists only have limited time to act as reviewers. The consequence is that journals are tending more towards a straight forward acceptance or rejection, rather than longer supported revisions processes. Over the 20+ years I’ve supervised students, some have published several peer-reviewed papers whilst others have not managed to get one accepted. I never achieved a first-authored one during for my own PhD.

I think the differentiating success factors in getting publications are writing up research that is novel (rather than incremental), maintaining self-belief in your work plus a small measure of luck. Many times has a paper been rejected, only to be accepted by another journal after revisions, but for a PhD student the rejection can be a very disheartening process; especially if it’s their first paper.

Therefore if you get rejected, don’t be down-hearted. Use the valuable reviewer feedback to look at the paper with fresh eyes, and give careful thought on where to submit. A lower-ranked journal can be better for a first PhD submission; especially if the research is still in the initial stages of development. Believe in your work, believe in yourself and send the paper out again and do this over and over until you get it accepted. You never know you might get three papers published like Sue!