Generate quick look of Sentinel-2 imagery

Hi all !

I have a dataset of Sentinel-2 images downloaded online(around 120 images). I want to pre-select images and exclude images with a high cloud cover before processing the dataset further. Does anyone know if there is a way to generate a quick look of the images in SNAP so I can then manually exclude unsuitable images without opening each image in SNAP ?

There is a tool called pconvert provided with snap.
You can use it to create quicklooks. You find a description in the help.

But maybe there is an easier way for you to do a preselection.
The file MTD_MSIL1C.xml contains a xml tag:

<Cloud_Coverage_Assessment>3.3353</Cloud_Coverage_Assessment>
It provides you the percentage of cloud coverage.
If you do your batch processing e.g. with Python then you can do the preselection on this value.

2 Likes

Hi marpet,

Thanks a lot for your response !! I really appreciate your help. I am processing images where I am interested in only a very small part (less than 5%) of the tile, therefore I didn’t apply the cloud cover percentage filter when downloading these images. However, I might rethink this approach just to cut back on my dataset a bit more and exclude images with more than 80% cloud cover or so.

Two more questions:

  • Is the help tab which you have screenshotted in the STEP forum or can it be found elsewhere ? I can’t seem to be able to locate it in the STEP forum.

  • Do you potentially know if there is a way to set up automatic quick look generation for my entire folder of images ? Changing the input for every single of my 140 or so images would obviously take ages. I wonder if I can somehow direct the command line to select my image folder and then loop through all files within that folder ?

Best regards and thanks again,
fknappe

The help screenshot is from SNAP. In SNAP go to the menu Help / Help Contents.

There is a guide on how to batch process data products with gpt.
Bulk Processing with GPT - SNAP - SNAP Wiki (atlassian.net)

The general process is the same. You need replace gpt by pconvert and adapt the parameters. Probably several other changes need to be made, I currently don’t see. Maybe this guide can be starting point for you.

Another simple way to create command line calls is Excel.
Copy the path to each file in separate rows and then combine this with the parameters.
In the last column you have then a list of strings. You can copy these into a text file, name it *.bat and execute it. Aa list of files names ca be retrieved in Windows ono the command line by calling:

dir /B > fileNames.txt
This will write all file names into the specified text file.

2 Likes

Hi marpet,

Thanks a lot for your reply ! I resolved the issue by displaying the paths of all files ith Rstudio and the copying the list of files into the pconvert input. I will need to do some more batch processing though so the resources you’ve send me will probably be really useful for this.

Thanks again for your help, it is really appreciated !!