Closed vvvvalvalval closed 1 year ago
Do you see the locations as points or areas? If area, we may want to add another column to TIMED_LOCATIONS_CSV
:
timed_location_id, x, y, r
Where r
is a search radius
@lautenberger I see the locations as points.
Excellent, that makes it much more straightforward algorithmically!
This is implemented in commit https://github.com/lautenberger/elmfire/commit/0ca65b18efb61cb4d4582dc6181b2c3f5d3049f9, closing this issue!
Problem statement
For some use cases, it is valuable to know when (if ever) each fire burns certain spatial locations.
Proposed solution
Example
elmfire.data
:TIMED_LOCATIONS_CSV
is an input CSV listing the points at which timings are required, in EPSG:5070 coordinates:The number of rows in this CSV will typically be much smaller than the number of pixels in the landscape, maybe one point every square kilometer or so.
timed_location_id
is an opaque id for each location, which ELMFIRE will convey unchanged to outputs, making it easier for callers to relate inputs to outputs. Ideally any data type should be allowed, but I guess it's fine to make it, say, a 64 bits integer if that makes the implementation easier for ELMFIRE.TIMED_LOCATIONS_EVENTS_CSV
is an output CSV emitted by ELMFIRE, describing which fire burned which location at what Time-of-Arrival:For example, the 2nd row says that fire number
icase=1791
burned locationtimed_location_id=42
at 21.12 simulated hours from ignition.Callers should not expect any guarantees on the order of the rows - let them sort them themselves if needed.