Hmmm, not entirely sure what you’ve done with this graph, but some of your numbers look significantly out to me

The x axis is latitude, the y axis is angle of elevation.

I plotted the angle required for a geostationary satellite at an altitude of 35 000 km in purple (I know that is very rounded).

Then I used the "sweep 5 degrees either side" to set the allowed tolerance to + or - 5 degrees, and plotted the angle required for a geostationary satellite plus 5 degrees in red, and the angle required - 5 degrees in black.

This gives us the range the stratollites would need to be at in order to meet the required tolerance of +- 5 degrees.

Plotted in blue are the angles required to reach these stratollites.

For the graph, I used a stratollite elevation of 4100 km as an example, as that worked well.

I started with one at the equator, tweaking the height until I got the best coverage, then cut it off as it passed the red line.

I then added another one, setup to match where the previous one stopped, and again and again, ending up with 4 stratollites.

How do you get 1 “stratellite” per degree latitude if they are 100km high?

Take the simplest possible case, with it directly overhead. You only need to move 1.7km for the angle to target to change 1 deg on flat ground. It’s a bit further at lower elevation angle. I make it 3.2km at an elevation angle of 42deg, which is about in the middle of the range.

So even with the generous accuracy requirement of +/- 1 deg, each “stratellite” could only cover an area on average about 6.5km across. (The observed change in elevation angle due to the real geometry over this distance is so small, we can effectively ignore it). All while somehow perfectly focusing their transmissions to give continuous coverage without overlapping, and all of them somehow keeping position.

I used a tolerance of 5 degrees.

Also, the distance per degree increases with increasing separation between the stratollite and the area it is targeting.

If the stratollite is angled at an angle of a from straight down to the start of its range, and has a range of b, then the distance covered is given by:

h*(tan(a+b)-tan(a))

For a stratollite at 100 km directly above the equator, with b=2 degrees (for + or - 1 degree), this would give a range of roughly 3.5 km.

But with b=10 degrees, it would give 17.63 km, which is over 0.1 degree of latitude.

But if we go much further away, to say 60 degrees from straight down, we get 101 km, or close to 1 degree latitude.

But over 1 degree latitude change, we would also have the position to the geostationary satellite change by over 1 degree (which would continue until it drops below the horizon). Putting in an extra degree gives us 117.2 km. If instead we put in the angle corresponding to the distance from the first step we get 115.8 km. Either way, over 1 degree.

So faking it with stratollites becomes easier the further away from the equator you are.

But even with a range of 1 degree, you need a ridiculous amount of them to cover the contiguous US.