Earthdata Forum

Hi, I am working with ECOSTRESS L3T Evapotranspiration (ECO_L3T_JET, version 002) data and have encountered a consistent issue with coarse grid-cell artifacts. The artifacts were found regardless of the platform I used to download the data.

Specifically, I have noticed 500 m–wide blocky patterns (see attached figure) in the ET variable (including ETinst, ETday and ETinstUncertainty etc.) that persist even after applying all available quality masks, including the cloud, water, and QA layers. The artifacts are spatially regular and appear to align with the ECOSTRESS tiling grid rather than with actual landscape features. May I ask for advice regarding:

1. Whether these coarse 500 m patterns are known artifacts with the ECOSTRESS L3/L4 products, and;
2. If there is any recommended approach to further filter or mask out such artifacts beyond the existing QC, cloud, and water layers.

Thank you very much for your time and I look forward to hearing any advice.

Hello @henrychyeung Apologies for the delay in response. We are discussing your question internally and will post here when we have an answer, as well as respond to your email. Thanks -- Danielle

Hello,
The STARS algorithm (input into ET) is a data fusion between the high-resolution HLS (harmonized Landsat Sentinel) and the daily VIIRS data, which can result in artifacts like these when there isn’t enough high-res data (e.g. if the last Landsat image was cloudy). These artifacts will be improved in Collection 3.
The best course of action to filter the artifacts is to refer to the ECO_L2T_STARS UQ layer to filter out the artifacts. It may not exactly be helpful if you are looking to filter out the granules prior to downloading but they should at least be able to filter out the impacted pixels.
Thanks,
LPDAAC