I am searching for an algorithm to make a shape file of every shadow in Landsat image (30 meters per pixel). I need to classify lakes at the image, but territory is very mountainous and shadows are classified as lakes.

Found some article but i haven't found this algorithm as a code.


It would be perfect if you give me a link to ready to use algorithm (as a code or executable programm). It would be even greater if this code will be made for Arcmap or ENVI.

  • Doesn't hillshading in ArcMap automatically detect shadows? If so, then extracting all zero values of the image's hillshade grid will solve your problem. – whuber Oct 22 '12 at 13:17
  • You can try to search for use of infrared and thermal bands, in some band ratio index. – nadya Oct 22 '12 at 22:48

Why not use the SRTM dataset to find out where you actually have the shadows? In the case you are referring to you might find that shadows are cast by clouds or other interference giving you a wrong (biased) basis for shadows in your landsat imagery.

Find the potential shadow areas by using SRTM and make a hillshade raster. Make sure the sun angle and time of day matches the Landsat image you are using. This will give you a "neutral" basis to evaluate which parts of your landsat image are shadows and not.

After having made the hillshade raster you can convert it to a shapefile.

| improve this answer | |
  • I think hillshade from SRTM is not the best way for detecting shadows. And he needs shadows for the PARTICULAR image, with all it's clouds at that moment. (More, SRTM is a digital elevation model, and cannot help to find shadows from buildings and other large artificial structures if present. And there can be some landscape changes if the time of SRTM and the image is very different). – nadya Oct 22 '12 at 22:36
  • Well, s/he is looking for ways of isolating water bodies. In addition the area is "very montainous" so I guess this would lead to buildings becoming relatively small. In my opinion artificial structures become irrelevant. Other relevant information would be area for the analysis as well as purpose. Further more by knowing where the shadows are at least there is a way of eliminating potential "lakes" by saying that shadows falling in non-flat-aras are definitively not lakes. But you are right, the analysis is more complex. – ragnvald Oct 22 '12 at 22:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.