From GIS to Remote Sensing: SCP Questions of This Month: April

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.

Question

Can we use scp file created for landsat 8 images to landsat 7 images for the same region?. It shows wavelength not matching. How to correct it? or do we need to make the classification again for landsat 7 images?

Answer

You should use the button "calculate spectral signatures of highlighted ROIs" to calculate the signatures after the Band set creation (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/scp_dock.html#roi-signature-list)

Question

After SCP preprocess Sentinel-2 all the bands have 10m resolution, also the originally 20m bands? Is it an image fusion?

Answer

It is not an image fusion, 20m bands are simply rescaled to 10m resolution. Of course, the information of those bands still have 20m resolution.
You can use only 10m bands for classification, but generally using the other bands can provide more spectral information and improve the classification results.

Question

Now we are using the plugin version 5.3.6.1 but we are having some troubles with the classification results. Even though we are sure the new version of the plugin must be even better than the previous one, we are not able yet to addapt to the new interface.
If you could give us this previous tool we would be really gratefull.
If not possible , are there some tutorial videos of have to use it?

Answer

You can download any previous version of SCP from http://plugins.qgis.org/plugins/SemiAutomaticClassificationPlugin/
About the tutorials, please read http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/tutorial_1.html

Question

When we are using same training input file for satellite images over a region for different years, we recalculate the signatures and check the preview. But, when we recalculate, some class overlap happens some times, then we adjust some thresholds to remove the overlap. Will this affect the comparison of the classification across years? or do we need to check the class ranges every time and keep them consistent always.
How does the spectral information is stored in the classification scp file? Is it pixels's spectral information or spectral information of a particular category or both.

Answer

Spectral information is stored for each spectral signature as mean, minimum and maximum values for every band and the covariance matrix.
The comparison results depend on the classification algorithm. In some cases you may need to create new ROI polygons because of Land Cover changes. If you calculate the spectral signatures on different images you should check that ROI pixels belong to the correct class.

Question

I'm just switching from ArcGIS to QGIS. I regularly classify images for hydrological studies (I switch bands for near infrared imagery). For urban areas, I need the resolution of the 1m NAIP imagery, but I have only found that available in 4 band jp2. I am not seeing how to use jp2 in SCP. Can someone confirm whether it can be, and if so, how?

Answer

Yes, if you can open the image in QGIS, then you can use it in SCP. You can also use this tool to split raster bands https://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#split-raster-bands

Question

I'm just switching from ArcGIS to QGIS. I regularly classify images for hydrological studies (I switch bands for near infrared imagery). For urban areas, I need the resolution of the 1m NAIP imagery, but I have only found that available in 4 band jp2. I am not seeing how to use jp2 in SCP. Can someone confirm whether it can be, and if so, how?

Answer

Question

Just noticed that when downloading Sen2 through SCP we got 02 folders for each scene, one contains JP2 and other has TIFFs. I suppose that we use only TIFFs, so for what we need the JP2 ?

Answer

The jp2 files are the original images that are downloaded. You can delete them after the preprocessing.

Question

I can't find the ROI Creation dock in SCP Toolbar.

Answer

The ROI Creation dock was in an old plugin version. Please read this tutorial about the latest interface http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/tutorial_1.html

Question

I would like to know about the bands ratio and I want to classify water , forest , built up , and bare soil.

Answer

Please read http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/remote_sensing.html#spectral-indices and I suggest http://www.jars1974.net/pdf/rsnote_e.html

Question

Is there also a way to export the ROI extent as vector file (e.g. Polygon Shapefiles?)

Answer

Please see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#export-signatures

Question

I've just installed Semi Automatic Classification plugin on a new laptop with QGIS 2.18.6 (it works fine on my other laptop with an older version of QGIS).
Every time I start QGIS on my new laptop I get the error message:
"Please restart QGIS for executing the Semi Automatic Classification Plugin." I have tried uninstalling the plugin via the plugins menu, but that just makes Qgis Crash and gives me the "minidump" error again. This means I can't remove the plugin.

Answer

It seems a QGIS issue. Please install the QGIS version 2.14 Long term release

For any comment or question, join the Facebook group and the Google+ Community about the Semi-Automatic Classification Plugin.