Support efficient large Wildlife Recorder imports

One issue with very large Wildlife Recorder (and other format) imports is that they need geocoding en masse.

Simply scrolling through 1000 locations is unbearable; and even worse, it’s expensive for me - using the Google geocoding API for 1000 locations costs $5, and that’s a lot of cost to pay for a single user (who might repeatedly import, which is getting worse!)

Idea: for large imports (say, anything with > 10 locations, and where those locations all have place names - true for Wildlife Recorder, but not for Birdlasser), use eBird geocoding to:

Find a state
Download checklists from the state from eBird
Use those checklists to compute some sort of polygon shape (convex hull?!? though that might go very poorly for some states)
Find all points that line within that polygon, and map them all to the state automatically
Proceed to the first point that wasn’t mapped, and repeat.

Some implementation thoughts:

This doesn’t get US counties, but not sure I particularly care for very large sets of locations. Alternatively, I could do counties for 10 locations, and states for 100 locations, or something like that
Also, I imagine there’s some things which are states but have very very few hotspots in eBird; at some point, falling back to the country is probably a good thing.

‌

Comments (5)