Skip to main content

Hi, 

Hope you are doing well. I have downloaded nearly 50 tiles of the wet tropical rainforest of Australia, and a few tiles have severe cloud cover even after applying the filter function of Planet Explorer.  So, how can I remove those clouds? Unfortunately, there are no reference images available without Cloud for those locations in near times (2019). I am using QGIS. Any suggestions in this matter are highly appreciated. 

Kind regards,

Shawkat

Hi Shawkat, 

If you downloaded an analytic asset the tiles should have come with a udm file. If you downloaded a visual asset (you can tell which you downloaded from the file name) then you will need to reorder the imagery with the analytic asset. The UDM file should have 8 bands that each represent a different filter layer as you can read more about here. You will be interested in the first band, which is the clear band. 

Now if you load in the imagery and its corresponding udm file you can use the raster calculator to multiply the clear band (band 1) with each of the bands of you image. The output will be 4 separate clipped bands that you can restack using build a virtual raster and exporting that raster.

I hope this helps,

Octave


Hi Octave@ Thank you so much for the suggestions. I will try it. If there is any issues, I will contact you. 

Again many thanks.

Kind regards,

Shawkat


Hi Octave,

I checked the downloaded surface reflectance RapidEye ortho analytic_sr zip file (not the visual asset optimized for visual analysis-RGB only) through Planet Explorer. They have a UDM tiff file containing only one band [band 1 Gray]. I probably need to download the visual asset. I will try this. However, can we do something with the UDM file with one band downloaded from the surface reflectance asset?

Thank you.


Hi Shawkat, 

I am sorry that I led you astray. Any items captured from the RapidEye constellation do not have a UDM2 cloud mask, they only have an UDM (unusable data mask) which flags faulty imagery and clouds. The algorithm is a bit less robust than the UDM2 algorithm that is applied to all Planet Scope products. You can read more about both in this spec document. The gist is that it is a single band item where each pixel that has a value of 1 when there is assumed to be a cloud. I would try following the about above steps with a filter for the UDM pixel values.

Best,

Octave


Reply