Slight spatial misalignment in PlanetScope/RapidEye imagery

Question

Hello!I have a set of scenes of the same area, but there’s some slight misalignment between them (like 3 or 4 pixels), this usually isnt a problem, but I need the most precission possible for my use case (measuring coastal recession), so even a small misalaignment could change results greatly. This happens between images from different and same sensors, so it isnt a result of using different sensors (at least i dont think so),beacuse i have some that are correctly aligned. Also, i used the same AIO to download the images from PlanetScope, so the images have the same extent and are clipped to the same area.  My question is, is there any way to download scenes that are correctly aligned, or should I fix it manually? (or is there a faster way to automatically align them?)  Thanks everyone!

Mariana Curdoglo · Accepted Answer

Hello! My name is Mariana Curdoglo and I am the Product Manager of PlanetScope images. The product specification for PlanetScope images and RapidEye positional accuracy is 10 m RMSE for 90% of cloud free images. If the images you are using are more cloudy the positional accuracy can be worse, as you have observed. We did have some engineering publish the following blog post to help with these type of issues in the past: How to Co-Register Temporal Stacks of Satellite Images.

The only filter option you have is on ground_control true. For an image to be ground_control true, it would need to have gotten at least 200 ground control points. Otherwise, it would be marked and ground_control false. You can confirm in the metadata if they are true or false and try to stick to true only.

Mariana

AAPV · Answer

Hello! My name is Mariana Curdoglo and I am the Product Manager of PlanetScope images. The product specification for PlanetScope images and RapidEye positional accuracy is 10 m RMSE for 90% of cloud free images. If the images you are using are more cloudy the positional accuracy can be worse, as you have observed. We did have some engineering publish the following blog post to help with these type of issues in the past: How to Co-Register Temporal Stacks of Satellite Images.

The only filter option you have is on ground_control true. For an image to be ground_control true, it would need to have gotten at least 200 ground control points. Otherwise, it would be marked and ground_control false. You can confirm in the metadata if they are true or false and try to stick to true only.

Mariana

This was it, thank you Mariana!

Reply

Sign up.

Login with Planet Account

Login to the Planet community

Login with Planet Account

Scanning file for viruses.

This file cannot be downloaded