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 have problem on ground truth sampling for accuracy assessment. How can I choose a good sampling for it and how can I create a good sampling?
Answer
In general point sampling should be designed considering the land cover composition, in order to get accurate validation.
If the land cover is composed of classes with small area you should consider a stratified random sampling approach to get a minimum number of samples per class, but this is not provided in SCP.
Alternatively you can create random points (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#multiple-roi-creation) with the option "inside grid", and selecting "min distance" of a proper size depending of pixel size, then manually classify samples (see https://fromgistors.blogspot.com/2014/09/accuracy-assessment-using-random-points.html); this way the number of samples could be larger than the number of samples with the stratification approach.
If you know Python you could create the batch text itereating through the image files.
If the land cover is composed of classes with small area you should consider a stratified random sampling approach to get a minimum number of samples per class, but this is not provided in SCP.
Alternatively you can create random points (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#multiple-roi-creation) with the option "inside grid", and selecting "min distance" of a proper size depending of pixel size, then manually classify samples (see https://fromgistors.blogspot.com/2014/09/accuracy-assessment-using-random-points.html); this way the number of samples could be larger than the number of samples with the stratification approach.
Question
How do we get rid of the black area that appeared after clipping? Otherwise, the area would be included as unclassified.
Answer
You can set the color white or transparent for the NoData values. If you click with the Identify tool over a black pixel you can identify that value.
Question
How can i classify a Landsat mosaic?
Answer
You should first perform the DOS1 correction of single images. Eventually mosaic the bands. However, I recommend classifying the single images (you can use the same spectral signatures), and then mosaic the classifications.
Question
I want to loop the atmospheric correction over about 2000 images. Is it possible to access the Semi-Automatic Classification Plugin via the Python console in QGIS? Here's my batch code for converting one image:
sentinel_conversion;input_dir : '~path to files'';mtd_safl1c_file_path : '';apply_dos1 : 1;use_nodata : 1;nodata_value : 0;create_bandset : 1;output_dir : '~path to save files'
Answer
You should use the batch tool http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#batch for that. You should add one conversion line for each image as you did.Question
Clip multiple rasters tool doesn't run correctly on my computer. I work with Ubuntu 16.04 and Qgis 2.18. I follow the plugin instructions to install in ubuntu but when I run this tool with layers from landsat 8, Qgis close and don’t execute the clip. I’ve tryed purging Qgis and plugin several times, changing the location and name of the raster archives, using a shape archive to clip... without success.
Answer
I think it is related to QGIS installation and plugin dependency. You can try to install it again. If you still get the same issue, you could consider this virtual machine http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/semi-automatic_os.html
For any comment or question, join the Facebook group and the Google+ Community about the Semi-Automatic Classification Plugin.