Skip to main content

Hi,

 

I'm currently working on estimating chlorophyll-a (chl-a) concentrations in large water bodies using PlanetScope raw data, enhanced with acolite atmospheric correction. However, I've encountered a consistent issue where some scenes exhibit unusual lines or stripes, which I suspect to be noise. This has happened across multiple data sets and seems to disrupt the accuracy of my chl-a concentration readings.

 

Upon reviewing the PlanetScope data documentation (https://developers.planet.com/docs/apis/data/sensors/#the-psbsd-instrument), I understand that each frame consists of eight stripes. I believe these are causing the discrepancies in chl-a concentrations, as values significantly differ from one side of a stripe to the other. This variance is impacting the trustworthiness of my output.
 

Here's an image for reference

 

Hello @rmant 

Based on the information in this post, we believe support will be able to help you with this issue. We have created a support ticket on your behalf and have copied the email you have used in your Community account.


@elyhienrich can I also hear the response to this? I believe I have encountered the same issue.


Hi @Coastal Carbon 

I will update this thread as soon as I have any information :) 


Hi @elyhienrich , is there any updates with respect to this problem? We’re still facing this issue.


Hi @rmant 

Thank you for following up. Sorry to hear you are still facing this issue. Please reply to your open support ticket as the support team will be able to provide you with any updates. 


Hello!

My name is Mariana, and I am the Product Manager for PlanetScope images. The effect you are observing is tap imbalance. Tap Imbalance is reported in the Quarterly L1 Data Quality Reports. In homogenous areas, like water, we see subtle differences in numerical values that produce this artifact. If you look at the actual pixel values the change is very minimal in radiance data but because it is exactly linear it becomes very apparent visually between the taps. This is an inherent artifact to many remote sensing sensors, including the SuperDove sensor, and can be resolved on the post processing level instead of upstream on the imagery side. Usually this is done by applying a normalization or smoothness methodology that you are comfortable with. These methods tend to use some sort of spatial averaging of pixels.

Mariana


Reply