In my classified mangrove area there are small patches of oil palm misclassified as mangrove, does any one knows how to separate mangrove and oil palm in Landsat imagery?
It is easy only by ground survey you can save more than 100 points for mangrove tree area using GPS the add these 100 points e.g. to the Landsat imagery (all pixel into these 100 points is a mangrove tree).
If your study area is small 10 +- points
What`s the problem? You seem to know the oil palm patches.
What I would do assuming you applied a MLC? Mask the the known categories, and classify the remaining portions of the Landsat frame.
Did you try on masks? (Reclasss) or pass it to vector and reclassify?
Thank you all for your suggestion
Hein Van Gils and Eduardo Ramirez
We will test machine learning classifiers in the end instead of MLC.
This moment, I am interested in extracting only mangrove forest extent from other landuse and I am using combined vegetation index and DEM approach. I am using auto threshold based binary images from NDVI, NDWI and DEM ( to identify lowlands). This approach allows separation of mangrove from other landuse classes, but the problem with oil palm tree those are very close to mangrove forest are still unable to separate from mangrove (Attached figure).
It seems that the reflectance values are very close to each other and I would like to know if there is any way to identify these small patches of oil palm trees using any specific spectral signature?
- Badges
- Science topic
- Similar topics
- Geoscience
- Remote Sensing