search

nasaharvest / crop-mask

End-to-end workflow for generating high resolution cropland maps
Apache License 2.0
94 stars 26 forks source link

Visual assessment notebook #354

Closed ivanzvonkov closed 6 months ago

ivanzvonkov commented 6 months ago

Notebook for generating a script to do a visual assessment of existing maps + link to Senegal assessment. Also updates ASAP map based on prior discussion

Direct link to notebook: https://github.com/nasaharvest/crop-mask/blob/visual-assessment/maps/Senegal_2022/1_visual_assessment_existing.ipynb

review-notebook-app[bot] commented 6 months ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.

Powered by ReviewNB

hannah-rae commented 6 months ago

This is really cool! I like it. A couple of questions:

ivanzvonkov commented 6 months ago

Do you think we need to visualize all of them? It might help reduce the decision/exploration space to remove some of them that we know don't perform well. For example, Globcover tends to be pretty bad (that may be the only one I'd remove).

Yes, I agree visualizing fewer would be better. In the next comparison notebook I'll skip over Globcover. For Senegal I think it can serve a demo purpose of showing how bad some maps are 😄 so I'll leave it here for now.

Do you think it would help to visualize the consensus layer, i.e. the number of maps that predict crop in each pixel? (basically the sum of all of the maps) That way we can see where there is high agreement and disagreement, which could be helpful for the sample design.

Yes it would definitely be helpful, will work on this.