SCP Questions of This Month: August

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

I want to produce a number of change detection maps  for each class instead of produce a global change detection map for two classified images in different dates.
How can i get like these maps in QGIS?

Answer

Please use this tool http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#land-cover-change to create a land cover change raster, and then change the raster style in QGIS to display only selected classes of changes

Question

I have an Orthomosaic (Geotiff, Lat/Lon, WG84) from Drone flights , and want to perform an Unsupervised and Supervised classification. The image has 4 bands R-G-B and a fourth band which is Alpha band. Can this be processed in SCP ?

Answer

If your images have the near infrared band then you can use SCP. Unfortunately the alpha band doesn't include any spectral information.

Question

How do we choose to use Land Cover Signature classification or other algorithms for classification?

Answer

LCS can be useful when you have difficulties in separating similar land cover classes, because it allows for the editing of spectral signatures, and you must know the land cover classes a priori.
The other semi-automatic algorithms are based on distance or probability and can be used when you don't have good a priori information about the study area and land cover classes.

Question

I can't download Landsat images from SCP. Why?

Answer

It is still possible to download images, with the exception of Landsat images acquired after 30/04/2017 (because the
 new Landsat Collection (https://landsat.usgs.gov/landsat-collections) is not available for download from Earthdata (https://search.earthdata.nasa.gov/search/granules?p=C185174181-USGS_EROS&m=-43.03125!62.015625!0!1!0!0%2C2&tl=1487189304!4!!&q=Landsat&ok=Landsat)

Question

I am trying to run an accuracy assessment and do not see the ROI Dock where I can create a new shapefile and proceed with making multiple random points to use in my assessment.

Answer

The interface has changed. You need to create a training input first. Please read http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#multiple-roi-creation

Question

My purpose is to refine a classification from the previous one in order to classify unclassified pixels while keeping some pixels as they are classified.
How can I define an expression that specifies: anything within this polygon that is not classified becomes "X"?

Answer

You can use this tool Edit raster http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#edit-raster
You can set an expression for changing only value 0 (unclassified) under a vector polygon.
Please consider that this tool modifies the original raster directly, so you should make a copy of the classification before using it.
For instance read this tutorial https://fromgistors.blogspot.com/2017/01/direct-editing-of-raster.html

Question

Surface reflectance is obtained when we perform the atmospheric correction?

Answer

Yes in general surface reflectance is the actual reflectance at soil. Atmospheric correction is required, but also other corrections (e.g. topographic correction) may be required depending on the study area, and field measures could be useful for comparing image value with ground truth.

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