Hi,
I’m trying to use Idepix tool to classify my Landsat 8 data. When I try to run the tool, I get an error: org.esa.snap.core.gpf.OperatorException: Selected cloud screening algorithm cannot be used with given input product.
I guess this is due to some metadata missing. What I already performed with the images and what I did to import it is:
Download Landsat 8 Surface Reflectance data from EarthExplorer
copy it using arcpy with Snapping option (because I work with older Landsats also)
open each band in SNAP
combine them using Radar -> Coregistration -> Create Stack
create RGB composite
And Idepix won’t work with the above. If this is due to the lack of metadata, how can I repair this? I’m not able to work with fresh pure USGS scenes, because I need to preprocess it.
If this is because Idepix only works with Landsat 8 OLI/TIRS, not SR, then can I do anything? Is there a way to “trick” the tool? I need SR data, not regular images.
I will appreciate any help . Thank you in advance.
It is currently not possible to use the SR data with IDEPIX.
It expects TOA data. In the background a neural net is used and it is trained with TOA data. So, no way to trick IDEPIX here. Also the cirrus band is needed which is not available in the SR data, as far as I know.
In general the SR data format of Landsat-8 is not yet supported by SNAP. That’s why you had to open the bands separately. You should know that you miss important information which is written in the xml file.
For example the scaling factor. I think this is 0.000100. So you have to multiply each band value with this factor to get the real SR value.
Using the radar coregistration might work, but there is also a generic one. At Raster -> Geometric Operation -> Collocation.
If you want to use a classification you can either use the one provided with the Landsat-8 SR data. The cfmask band has the classification of clear, water, cloud_shadow, snow and cloud.
Or you can download the corresponding L1 data and process this with IDEPIX and then merge both.