From GIS to Remote Sensing
This post is a collection of questions and answers about the Semi-Automatic Classification Plugin (SCP) and remote sensing which were discussed in the Facebook group and the Google+ Community this month.
These questions vary from supervised classification technique to software issues, and can be useful to the readers of this blog for solving issues about the use of SCP.
These questions vary from supervised classification technique to software issues, and can be useful to the readers of this blog for solving issues about the use of SCP.
Question
I want to know if it is possible to have a report for each feature of a polygon layer overlaid in the land cover raster (as does the zonal statistics plugins).
Answer
Yes, you can use the tool Cross classification http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#cross-classification
Question
We just purchased RapidEye for Jan 2013 and Dec 2015. We want to do change analysis. Could you give me instructions on how to do land cover classification with RapidEye?.
Answer
For the classification you can follow this tutorial http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/tutorial_2.html from the point "Create the Band Set".
Atmospheric correction is recommended, but unfortunately SCP doesn't include tools for the preprocessing of RapidEye images. You can use the tool "Band calc" for performing the raster calculation for converting DN to reflectance values.
After the classification of both images you can perform Land Cover Change (see for instance this old tutorial https://fromgistors.blogspot.com/2014/09/monitoring-changes-in-amazon-rainforest.html).
Of course, considering the very high spatial resolution, you could consider Object Oriented classification algorithms, but these are not included in SCP (you can use Orfeo Toolbox for that).
Atmospheric correction is recommended, but unfortunately SCP doesn't include tools for the preprocessing of RapidEye images. You can use the tool "Band calc" for performing the raster calculation for converting DN to reflectance values.
After the classification of both images you can perform Land Cover Change (see for instance this old tutorial https://fromgistors.blogspot.com/2014/09/monitoring-changes-in-amazon-rainforest.html).
Of course, considering the very high spatial resolution, you could consider Object Oriented classification algorithms, but these are not included in SCP (you can use Orfeo Toolbox for that).
Question
Is it possible to classify something with *only* its spectral signature or band ratios? For example, a mineral like Quartz. I do not know which pixels in the image are Quartz so I can not select them as ROI's/Training Areas. But I have the spectral signatures imported from the USGS library.
Answer
Yes, it is possible. You should first create the Band set with your image (for instance Landsat) and the training input, then import the spectral library (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#download-usgs-spectral-library ).
Question
I've stacked 10 Bands of Sentinel image and whenever I try to plot spectral signature, the lines gather on a single band as you can see in the image. Why?
Answer
The problem is related to image values, which are too large compared to the wavelength. You should preprocess the bands dividing by 10000, or set the Multiplicative Factor in tab Band set http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#band-set-definition
Question
In the calculation of Band Set the maximum and minimum values are identical, however, when I check the individual values of the pixels, directly in the raster, the values are different.
The raster result of the Band calc calculation is totally black.
In the calculation of the NDVI, a difference in the maximum and minimum values of the pixels is noticed. The values of the individual pixels are also different, the mean, the standard deviation, all different when comparing the results between the Band calc and QGIS Raster calculator.
However, in the NDVI calculation, it is generated raster with normal appearance.
The Band calc calculation, the generated raster, is all black.
Answer
The fact that the raster is black should be only related to the style single band gray.
You should be able to assign a style to the raster. Pixel values should be the same as the raster calculator. See for instance this tutorial https://www.youtube.com/watch?v=a8ulRtzLjgk
You should be able to assign a style to the raster. Pixel values should be the same as the raster calculator. See for instance this tutorial https://www.youtube.com/watch?v=a8ulRtzLjgk
Question
We can login and log out without any problems on Sentinel data hub.
When we tried to download Sentinel images in SCP we got a window to log in.
Answer
Usually it takes some time before new users can access the API Hub (used by SCP). You can try again in a few days.
Question
I would like to do both a DOS1 correction and pan-sharpening of my Landsat 8 images. I've tried checking both boxes in the pre-processor, but then I get two separate files for each band - one that is DOS1 corrected and one that is pan-sharpened.
If I do only pan-sharpening, the image gets darker (as if DOS1 has been done too).
To perform both, must I first perform DOS1, and then afterwards perform pan-sharpening?
Answer
If you check both DOS1 and pan-sharpening, first DOS1 is performed, then pan-sharpening is performed using DOS1 corrected bands.
For any comment or question, join the Facebook group and the Google+ Community about the Semi-Automatic Classification Plugin.