Hello,
I am currently working on extracting high-quality Leaf Area Index (LAI) from Sentinel-2 L2A products. My input data are L2A products generated by the Theia Land Data Centre (MAJA algorithm), with a muscate format output. So my point is not to report an issue, but to get your help on the best practices, as we’re going to industrialize this workflow.
My current simplified workflow in SNAP is:
-
Read
MTD_ALL.xml -
Resample
-
Biophysical Processor
I succeded in generating a LAI file, but to ensure the best scientific quality of the output, I would appreciate your advice on a few specific parameters:
1. Resampling strategy: What is the best practice for the resampling step before the Biophysical Processor? In the resampling parameters, by default it’s using Surface_Reflectance_B2 as a reference band, should I let it like that? And what does it mean? It will resample all the bands at the same resolution as B2 (10m)? For Upsampling method, by default it’s the Nearest method (and Downsampling is at “First”), should I let it like that?
2. Choice of the Processor: I’m currently using Optical > Thematic Land Processing > Biophysical Processor (LAI, fAPAR…) > Biophysical Processor S2_10m. Should I use the standard BiophysicalOp or the Biophysical10mOp? How do they differ regarding the input requirements and the output accuracy?
3. General Pre/Post-processing: Are there any other recommendations for pre-processing or post-processing to ensure the LAI values are optimal?
Thank you very much in advance for your time and expertise!
Best regards,
Clément AOUIZERATE