Line of sight analysis for entire US using Living Atlas elevation layer?

[Jay_Gregory]

I trying to do some line of sight analysis.  I have two point datasets, and need to figure out which points from dataset a (elevation of 30 f AGL) can be seen from dataset b (elevation of 2500 agl).  Is there any wholesale way to do this in Pro using a publicly available elevation service?  I was using linear line of sight in the defense toolbox, but I think the processing extent of the elevation service (5000x5000) prevents large scale analysis. 

If not, is there a way to reasonably break up this analysis into smaller jobs to achieve this?

Any tips would be helpful!

[BobBooth1]

I think I'd try buffering the 2500 feet points by whatever the distance to horizon is at that height (several miles?) and then use those (or use a "Select by Location" with that distance, probably faster) to select the points within that distance for analysis.

You could iteratively process them using ModelBuilder or Python.

[Jay_Gregory]

@BobBooth1 thanks for the response.  Does this take into account the mountainous regions? I would imaging an observer at 2500 feet (a plane) the horizon would be a lot closer in the face of the western mountains.  Your process doesn't even account for the terrain/DEM does it?

[BobBooth1]

Jay,

No, I was just thinking of it as a way to partition the data, so you're only considering the input target points that are within the horizon, limiting the amount of DEM you have to get from the service for each iteration.

You could get fancier, by sampling the ground elevation layer at each observer point and adding that to 2500, and calculating what the horizon distance was for that point, and buffering accordingly.

[Jay_Gregory]

Ah I see - this is interesting.  But wouldn't the distance to the horizon change based on nearby terrain.  On a flat featureless earth it is (according to AI) about 53 nm, but if a nearby mountain is 10k feet, and you are at 1k feet, I can be as far as 140 nm away (also according to AI) and still see the top of the mountain.  

[BobBooth1]

Hm. Yeah, I agree, spots on mountains are likely to be visible from further away.

Depending on how your data is distributed, you could use the simplifying assumption of a standard, viewer-elevation dependent horizon zone for some of it and, and then go back and do extra processing of points on mountainous terrain.