[nozzlegear]

This is really neat. I'm curious where the data for the tree shadows comes from though. I was surprised to see that the trees in my yard and my neighbor's yard were all mapped by your service, since I live in a small town in the middle of nowhere. I read the "how it works" FAQ section, which explained that building shadows come from the map services, but it didn't mention trees.

[isoprophlex]

I built a similar shadow mapping tool for some commercial party that wanted to accurately estimate solar panel production in The Netherlands... In my specific case I could access very accurate LIDAR heightmaps gathered from planes.

This means you can ray-march the location of the sun throughout the year over the entire country to calculate exactly where and when a surface is occluded by shadows from nearby (or even faraway, sometimes) objects.

The LIDAR data can be as detailed as a shadow cast by antennas, a chimney or a tree... Which is more important than you'd think, because a little bit of shadow on a single panel means that all panels daisy-chained to that panel will see an efficiency drop! (So you either don't chain them but give each panel its own inverter, or you wreck your neighbors chimney)

[tppiotrowski]

You can train ML to recognize vegetation in satellite photos and further train it to estimate its height +/-3 meters

https://github.com/facebookresearch/HighResCanopyHeight