0

I recently got some 4band images from Planet. When I attempted to calculate NDVI, the results seemed off. Once i got to looking at the bands, I found that the NIR band and the visible light bands were misaligned. I purchased the PSScene images from their analytic_udm2 bundle. Bundle descriptions here. As far as I can tell, these were meant to be orthorectified. However, when I compare red to NIR, I find that they are misaligned. They are off by about 100 meters. These are 2 bands in a single raster block.

enter image description here

In this image, red is shown as red and NIR is shown as green.

When comparing the raster to their publically available NICFI basemaps, I find that only the NIR band aligns with their analytic basemap.

enter image description here

Also with their visual basemap: NIR aligns, the visual bands do not:

enter image description here

¿Is there a way to fix this so I can get my NDVI calculations?

I apologize if I am missing something obvious, I am relatively new to this.

All images shown as examples are not analyzed at all but directly put into qgis from download and visualized using multiband color.

Improve this question
6
  • If I'm not mistaken, the UDM2 product is usable data masks. Not the red and NIR bands.
    – Micha
    Commented Feb 17, 2023 at 20:10
  • To help further, we would need: 1- the exact analysis you are doing in R; and 2- a sample (clip) of the original planet imagery.
    – Micha
    Commented Feb 17, 2023 at 20:11
  • Am wondering why you are asking this question here and not Planet directly? This is certainly a Planet issue either in their otrhorectification or pansharpening pipeline.
    – Jeffrey Evans
    Commented Feb 17, 2023 at 20:31
  • The analytic_udm2 product includes several items as indicated by the link in the question. It has the visual image, the analytic raster and a UDM. I am trying to get basic NVDI (NIR-R)/(NIR+R). As mentioned at the end of the post, the images I am showing are original imagery, unmodified. I simply colored them to show misalignment. I am asking here because I did not know if this was a normal thing and maybe it was a simple step to fix it myself or if there was no choice but to turn to Planet for a solution. As mentioned I am new at this and am not sure if this is a Planet issue or me.
    – Bushidoenator
    Commented Feb 17, 2023 at 21:28
  • I think we'd need to see the coordinate system reference metadata for the images to have a chance of this. If its the same coordinate system, and from the same supplier, then I'd take it to them. Do either of them line up against OpenStreetMap or Google maps?
    – Spacedman
    Commented Feb 17, 2023 at 21:34

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.