Good day,
I want to ask about automatic forest classcification.
Is there any index to automatically extract forest areas in Landsat 8? Landsat TM ETM and relatively successful Forest Index (((b4-b3-0.01) / (b4-b3)) * ((1-b4) / (0.1 b2))) Landsat 8 is not very useful.
Thank you for your answers and contributions.
I also think downloading a classified land use map from USGS is your easiest option. However, you can also automate a forest classification using image processing software like ENVI (or even ArcGIS) by creating training samples of known forest locations and running a supervised classification algorithm.
This is a tutorial of classification procedures in ENVI:
https://www.harrisgeospatial.com/docs/Classification.html
And this is a tutorial for ArcGIS:
http://desktop.arcgis.com/en/arcmap/latest/extensions/spatial-analyst/image-classification/image-classification-using-spatial-analyst.htm
Hello, thank you very much for your answers.
I am working on threshold based classification. Although SVM and supervised classification are relatively good methods, it doesn't give the results I want, especially when separating the boundary pixels from each other.
Formulas for some of the indices I use;
([Mean NIR] - [Mean Red] -0.01) / ([Mean NIR] - [Mean Red])) * ((1 - [Mean NIR]) / (0.1+ [Mean Green]))
(0.9351332 * [Mean Red] - 0.1183029) / (0.9351332 * ([Mean NIR] - 0.1183029) + 0.08 * (1 + 0.9351332 ^ 2))
(sin (43.09407) * [Mean NIR]) - (cos (43.09407) * [Mean Red])
(([Mean NIR]) - (0.9351332 * [Mean Red]) - 0.01) / ([Mean NIR] - (0.9351332 * [Mean Red])) * ((1- [Mean NIR]) / (0.1+ [Mean Green]))
I want to discuss it. To learn whether there are any indexes in the classification of forest areas.
Perhaps this paper can help. https://www.tandfonline.com/doi/abs/10.1080/10042857.2004.10677383?journalCode=tpre20