Combining the height of 382km from the notebook with the supposed resolution of 20 cm from that link gives an angular resolution of 0.00003 degrees [1]. That is about 2.5 * the angular resolution of the Hubble telescope. However, as that article notes, this image was taken through atmosphere.

I don't know much about telescopes, but some googling seemed to reveal earth based telescopes with similar resolutions. Hence, it seems within the range of possibility.

[1] https://www.wolframalpha.com/input/?i=20cm+%2F+382+km+radian...

The interesting thing you can derive is what the lower bound is on the diameter of the imaging system via the diffraction limit. I get about 1 meter (but have fat fingers and sleep deprivation). Which is big, but not crazy big.

Edit. 224 is supposed to be a variation of this guy [1], which has a 2.4 metre mirror (which should be more like 7cm on the ground).

[1]https://en.wikipedia.org/wiki/KH-11_Kennen

Does that 7cm figure include the effects of atmospheric dispersion?

That is the minimum possible resolution* (hence the lower bound on the diameter). Atmospheric stuff will make it worse and be variable from day to day (and depends on the length of the exposure, which would be a bigger problem looking at stars than the earth). I think that is around 2 microradians for astronomical telescopes.

*there are tricks to get super-resolution, but ignore that stuff.

Lockheed famously hires a lot of adaptive optics people for "god knows what."

I think the answer is obvious now.

I want to add that usa-224 flies at a remarkably high altitude for a spy sat, and certainly higher than previous gen sats with perigees below 250km

From the wiki article:

orbital altitude of 250 km corresponds to a ground sample distance of 0.06 m (6 cm, 2.4 inches). Operational resolution should be worse due to effects of the atmospheric turbulence.[34] Astronomer Clifford Stoll estimates that such a telescope could resolve up to "a couple inches. Not quite good enough to recognize a face"