SCP Questions of This Month: December

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 don't understand why there is no Sentinel-2 data for Berlin from June 2016 onwards when there is lots on the science hub - the api should have it as well, shouldn`t it!? Or how can I access through the amazon cloud?

Answer

I think it is related to the maximum number of results and the date range.
If you increase the Results number (e.g. 100) in Search or set June 2016 in the Date from (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#search-sentinel) you should get the most recent images.

Question

I compared some NDVI products of the same scene/ same day (for s2a/landsat) with different sources or processing ways.
1) NDVI from "Landsat Higher-Level Data Products" which is resulting from the NASA produced SR.
2) NDVI resulting from your PlugIn after preprocessing of the scene
3) Sentinel 2 NDVI previously processed with Sen2Cor
I noticed that NDVI values from 1 are more similar to 3 than 2.
Since I think that the Sen2Cor has the best Atcor, I would trust the Product 3 the most and conclude that the product resulting from your plugin is better than from NASA product.
Anyway, what I want to ask:
Is there a document where I can read about what is actually happening in the "preprocessing" of the data?!

Answer

The plugin uses a simple atmospheric correction (DOS1, you can read about this here http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/remote_sensing.html#dos1-correction).
Surface reflectance products produced by NASA or Sen2Cor use advanced atmospheric correction which is more accurate than DOS1,
If you need accurate NDVI measures you should use those products (read more here https://fromgistors.blogspot.com/2015/01/landsat-8-surface-reflectance.html).

Question

I'm studying about soil moisture retrieval so I would like to ask that can we use the SCP for this purpose?

Answer

About soil moisture please see http://www2.le.ac.uk/departments/physics/research/eos/format-eo/final-group-presentations/group-4-soil-moisture-mapping
Unfortunately SCP doesn't have specific tools for soil moisture.

Question

I have 1 question for the pre process under QGIS 2.14.9 LTR(64bit) and SCP 5.2.4.
I tried to use Landsat 8 data. I set this image as an input image and then set training input.
But SCP said "Error[61]: Projection error. Try to assign projection to raster **"
Before this operation, I already set projection to the image, EPSG 3100 (JGD2000/UTM zone 54N).  I got these bands from USGS Earth Explorer, then merged in ArcGIS the tiff files into 1 file and clip them.
I was able to import and classify the image by SCP 4.9.6 or before...

Answer

Try to create the new geotiff data by QGIS again and change projection to JGD2000/UTM ZONE54N with QGIS.
I could set that geotiff data as an training input, correctly.
Also, you don't need to merge the bands in one file, you can simply select the bands in the Band set tab http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#band-set

Question

How can I calculate NDVI in Band calc especially for fire monitoring?

Answer

Question

I am trying to classify images from the same place at different time series. When i create training polygons it is necessary all of the images and bands to be loaded in the project? Do i have the potential to take samples from only one band or rgb and load the other images afterwards?

Answer

You can use the same polygons for different images, but you must calculate the spectral signatures again before the classification.
After the creation of a Band set. highlight the ROIs in the ROI signature list and click the button to calculate spectral signatures http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/scp_dock.html#training-input

Question

Do you think if I want to find out LST for my home city it will be easy as loading the right LANDSAT or ASTER image instead of the Paris?

Answer

You can use Landsat images for your area; however, for better accuracy the emissivity values should be calculated from field survey. Also, you can use NDVI.for emissivity estimation. For the methodology you can read this paper Sobrino, J.; Jiménez-Muñoz, J. C. & Paolini, L. 2004. Land surface temperature retrieval from LANDSAT TM 5 Remote Sensing of Environment, Elsevier, 90, 434-440


Question

How can I import the previous ROI file (.shp) in this latest version? Otherwise, I have to make new .scp file and again create new ROI..
Secondly, I do not want that my ROIs would automatically be saved in the signature list. In the previous version, I have had such choice and in this current version, I do not find it.

Answer

You can use this tool for importing ROIs http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#import-signatures
With this new version signatures are saved with ROIs in the same file. You can disable the automatic calculation of signature (Calculate sig.) in http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/scp_dock.html#roi-creation
You can use the button to calculate spectral signatures of highlighted ROIs in http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/scp_dock.html#roi-signature-list

Question

In ubuntu, I get the error message about missing dependencies, and to restart, but I can't uninstall the plugin, and QGIS crashes.

Answer



Please follow this installation guide http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/installation_ubuntu.html and install the missing dependencies (GDAL, OGR, NumPy, SciPy and Matplotlib).
You can also uninstall the plugin manually (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/faq.html#how-to-install-the-plugin-manually). If you still have issue with the installation, you can try the virtual machine based on Linux http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/semi-automatic_os.html

Question

I have a question about the pan sharpening; I have read the documentation and I know that the Brovey Transform is applied when the pan sharpen function is carried out when using SCP. My question is: the Intensity has the same value for all the calculated pan sharpened bands? the reason is that the pan sharpening works very well when is necessary to combinate bands. However, when I calculate an index (as NDVI), the result does not seem affected by pan sharpening, showing as apparently 30 m resolution image (although the file has resolution 15 m). 

Answer

You are welcome Luis Carlos Fernández García. About your question, as you can see here http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/remote_sensing.html#landsat-satellite, the Pan-sharpening process Brovey Transform MSpan=MS∗PAN/I (see http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/remote_sensing.html#pan-sharpening) is not effective for the NDVI calculation because substituting in NDVI equation (NIR-RED)/(NIR+RED) the factor PAN/I is common to all bands. Therefore you get the same NDVI as the 30m bands.

Question

I'm struggling with generating the final classified output; after clicking run, it comes up with error 28 (set ram to lower value) after a while. I have set it to the lowest value possible, but it still doesn't work.

Answer

It seems a RAM issue related to the 32bit system (64bit systems don't have this limit). You should try to set 256MB in the RAM value in Settings http://semiautomaticclassificationmanual-v5.readthedocs.io/en/latest/main_interface_window.html#processing or if you can install the 64bit version of QGIS.



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

Newer posts Older posts

Follow

RG