Back to DrivenData | Blog

Submission with additional bands: What are the corresponding paths?

@kwetstone: When preparing a main.py file for a submission which uses the two B01 and B11 additional bands, is it right to assume that the corresponding chips are located under /codeexecution/data/B01_images and /codeexecution/data/B11_images?

As far as I am aware, additional bands have to be downloaded through the Planetary Computing. I’ve tried it but without sucess, I always get a connection error.

Cheers.

@ials Per the problem description, only four bands are provided with the competition data in /codeexecution/data - B02, B03, B04, and B08. jsensio is correct, any additional bands have to be downloaded through the Planetary Computer. We’ve provided a tutorial showing how this can be done. Good luck!

Hi am using the code from the tutorial for downloading extra bands during the submission, but I get a connection error to the planetary computer. Have you tried and validated that it is working properly ? Thanks.

Thanks @kwetstone & @jsensio for your replies. I have used code from the tutorial for downloading additional bands when submitting. I´ve got this error message:
Running code submission with Python
+ conda run --no-capture-output -n condaenv python main.py
Traceback (most recent call last):
File “/codeexecution/main.py”, line 33, in
import rioxarray
ModuleNotFoundError: No module named ‘rioxarray’

Could you please install rioxarray in the codeexecution environment?

2 Likes

@ials thank you for pointing that out! We have added rioxarray to the code execution environment, and it should run successfully now. Apologies for the trouble.

Thank you very much, @kwetstone.

hi @ials , have you figured out what are the paths for additional bands?

I got permission errors when trying to pull and download the additional bands in the test feature chip folder.

OSError: [Errno 30] Read-only file system: ‘/codeexecution/data/test_features/aaaa/AOT.tif’

Thank.

1 Like

@hijunli: I understand that it is not allowed to save additional input features into the code execution environment. Thus, we need to pull down the additional bands and use them on the fly for prediction.

@ials Thank you! I will try to pull and calculate in memory.