Improving sampling - integrating areas of interest and different sampling strategies.

[philip-monk]

Would it be possible to implement a sampling notebook and/or add develop the existing geemap.extract_values_to_points API (or implement an additional API)?

Although this could be done in GIS (and that might be the best option - I don't know!), when I asked a question on the EE deeloper forum Ujaval Gandhi pointed me to this example - https://code.earthengine.google.com/f181880a1f0236206f75fa085c9b6c6f. (I assume the sample points are skewed relative to the polygon as it isn't a geodesic projection?).

Developing this further, would it be possible to:

• Vary the shape of the area of interest (e.g. circle, elipse, masked area, area s with specific qualities (e.g. height, temperature, land type).
• Vary the sampling strategy, e.g. probability-based (random, stratified, systematic), spatial (point, line, quatrat/area), and combinations thereof (giving nine principal approaches). I'm sure there are other strategies appropriate for RS data of which I am unaware.
• Accomodate/flag projection influences (e.g. in the example above).

(When extracting data I assume all data from each point is extracted, and not just the topmost 'visible' layer on screen?)

Thank you for your time.

[giswqs]

I have added a random_sampling() function. See the notebook example.

https://geemap.org/common/#geemap.common.random_sampling
https://geemap.org/notebooks/random_sampling