I am dealing with Landsat imageries. I am using ENVI for deriving my land cover. Water bodies are not among my features of interest and I need to identify them in order to mask them out from my imagery. This is to facilitate the process of image processing and preventing potential errors related to water bodies.
Do the band ratio to 7,5,4. Then the water bodies will appear in dark/ black color which helps in water body extraction.
You can cross check the extracted shape file later by using supervised classification process through erdas.
Dear Masoudi,
Before start your work apply an unsupervised classification in any platform (ENVI or other platforms) and classify your water bodies. Most of the unsupervised techniques can classify water bodies very well if your dataset contains SWIR wavebands like landsat dataset.
Best Regards
Thanks a lot Onur Satir for your input. Would you please give me your opinion on NWDI (Normalized Difference Water Index) as described in the paper whose link I am attaching to this reply. Thanks!
http://www.tandfonline.com/doi/abs/10.1080/01431169608948714#.VrCIh7J9670
I think NDWI could be a good a rapid option for masking water bodies. You can set a threshold for building a mask from this raster.
Another option in ENVI, which takes more time, is the spectral unmixing. You will generate a raster with pixels containing proportions of water responses and can create a mask via threshold, too.
Thanks everyone for the valuable inputs.
Based on the notions suggested in the papers whose links I am sharing below, I calculated NDWI using the GREEN and SWIR (the shorter one) bands. Using a spectral screening, I applied the threshold of -0.30 for masking the water bodies. The estimated extent for water bodies was quite reasonable (5.62%) compared to that of reference data, 5.61%.
PS: All the analysis was performed in ENVI. As to the satellite imagery, I used Landsat imageries. My imagery acquired in 2015 (Landsat 8).
http://www.tandfonline.com/doi/abs/10.1080/01431160600589179
http://www.ingentaconnect.com/content/asprs/pers/2009/00000075/00000011/art00004
You can use:
Feyisa, G.L., Meilby, H., Fensholt, R. and Proud, R.S. (2014). Automated Water Extraction Index: A new technique for surface water mapping using Landsat imagery. Remote Sensing of Environment, 140, 23-35.
USGS has started providing Quality Assessment (QA) bands with Level 1 and Level 2 Landsat scenes. This additional band is very helpful in identifying water pixels within a Landsat scene and subsequently masking them out. In addition, clouds, cloud shadow, snow/ice, clear terrain and terrain occlusion can also be identified with the QA band.
how I can water pixels mask out in Nbr index for assessing fire severity
- Badges
- Science topic
- Similar topics
- Biological Science
- Bioecology
- Systems Ecology
- Landscape Ecology