Estimating EECU hours
- Dominant language
- Python
- Stars
- 371
- Forks
- 48
- PR merge metrics
- No merged PRs in 30d
Description
This is definitely a nice-to-have, but I am wondering if there is a reliable way to estimate the number of eecu-hours. Considering we pay for these hours, it would be nice to avoid the "oops just spent 1k+$ on an experimental dataset" issue.
We currently export data to fixed-size tiles and take a sample of tiles, run an export task to cloud storage, and get summary stats of the eecu-hours with `ee.data.getOperation(f"projects/earthengine-legacy/operations/{task_id}")`. This allows us to roughly estimate the cost of "ingest".
I think this is hard in the general case, but maybe we could build a recipe to sample/slice in time/x/y/dims in order to build an estimate of eecu-cost? In reality, this would be a nice-to-have function on image collections themselves, but I am guessing `ee.data.getPixels`, export to cloud storage, or other options vary in eecu-time. Thoughts?
Contributor guide
Assessment
This issue has not been assessed yet.