Closed ranchodeluxe closed 1 month ago
I haven't spent much time looking into this but since I fixed the primarykey
to use data['t']
I'm starting to think that this logic might wrong: https://github.com/Earth-Information-System/fireatlas/blob/primarykeyv2/fireatlas/postprocess.py#L154-L158
And might explain why there are more geometry outputs for v3 snapshots
I haven't spent much time looking into this but since I fixed the
primarykey
to usedata['t']
I'm starting to think that this logic might wrong: https://github.com/Earth-Information-System/fireatlas/blob/primarykeyv2/fireatlas/postprocess.py#L154-L158And might explain why there are more geometry outputs for v3 snapshots
We're currently exporting dead=1
records too which accounts for the more verbose outputs, will fix that
The "weird" geometries we are seeing seem to relics of the previous hull fixes that stuck around and merged with newer static sourced fires
Problems
Comparing API outputs for v2 and v3 in the S3 corner of New Mexico shows some obtuse geometry outputs. And also a bit more noise in terms of number of geometries. Removing static sources is on for both runs
V3
API Browser
V2
API Browser