The photographs NASA released during Artemis II are the kind that stop you: auroras threading a paper-thin atmosphere, city lights tracing the coastlines, brown desert bleeding into green. Beautiful — and, for an architect, a distraction from the harder instrument. Behind the crew’s window sits a fleet of Earth-watchers, and the most interesting of them for anyone who builds with light is PACE — the Plankton, Aerosol, Cloud, ocean Ecosystem satellite, launched by NASA in February 2024.

PACE does not take pretty pictures. Its Ocean Color Instrument reads reflected sunlight in continuous hyperspectral steps of roughly five nanometres, from the ultraviolet through the near-infrared — hundreds of bands where an ordinary camera has three. That is the same move NASA’s SPHEREx observatory makes overhead: the Spectro-Photometer for the History of the Universe, launched 11 March 2025, reads the sky in 102 infrared colour bands, and on 15 April 2026 used them to map water ice and polycyclic aromatic hydrocarbons across the Cygnus X star-forming region. Stop treating a surface as red-green-blue and start treating it as a spectrum with a chemistry attached — the question the astronauts asked looking down (“is that haze dust or smoke?”) is answered not by the eye but by the band.

←TODAY: PACE resolves Earth’s reflected light in ~5 nm hyperspectral steps from low orbit, turning a colour into a chemistry.
→3012: A Zurich facade is read the same way — not as a render but as a spectral and solar fingerprint the building carries for its whole life.
Fulcrum: The coordinate transform that flattens a hyperspectral globe onto a flat map is the same one flattening your georeferenced site onto a plan — get the projection wrong and every angle downstream inherits the error.

Here is the part the press release skips. To place any of that data, NASA must solve the oldest problem in our trade: a round Earth onto a flat map. It is never free. The 3D geoid to the 2D projection always trades away either area, angle, or distance — you choose which lie to tell. When NASA’s Earth Science Division approved Tomorrow.io’s Ka-band radar data on 9 March 2026, through its Commercial Satellite Data Acquisition (CSDA) program, the headline number was a 0.98 geolocation correlation against a digital elevation model. That 0.98, not the imagery, is the load-bearing figure: it says where the pixel actually is on the ground. Its sibling numbers tell you which spacecraft to trust — R2 hit 0.93 correlation to ground radar with a −6% bias, while R1 managed only 0.73 with −22%. A survey-grade site model lives and dies on the same discipline.

And the Sun. Every latitude and longitude you build at sits under a specific solar geometry, set by the axial tilt of 23.44° and the slow wobble of the axis — the precession Aristotle’s inheritors called the Great Year. That full loop of the axis takes not a couple of thousand years but roughly 25,772 — awe, held under arithmetic. On the century your building must stand, the tilt is effectively fixed, which is exactly why the sun angle at your site is a hard number you can compute today, not a mood.

Atelier: An office adopting AI-driven site analysis this week should stop trusting a plugin’s default sun path and verify it against first principles. Monday move: pull your project’s true latitude from the georeferenced model and hand-check one solar altitude before you let any generative tool place a louvre or a PV array off it.

Hack: Pin the midday sun to your latitude before you draw a single shading fin. Solar declination swings with the day of the year; noon altitude is just latitude minus declination. Run this for Zurich (47.37°N) and read the angle your south facade actually faces at summer solstice.

import math
lat, doy = 47.37, 172
decl = -23.44 * math.cos(math.radians(360/365 * (doy + 10)))
print(round(90 - abs(lat - decl), 1))  # noon sun altitude, deg

You get about 66° — the ceiling your overhangs must clear in June. Change doy to 355 and watch the winter sun drop to roughly 19°, low enough to walk straight under the same fin. That gap between 66 and 19 is your whole passive-shading strategy, and it was decided by orbital mechanics long before it was decided by you. Compute it once, from the number, and let the frontier instruments overhead confirm what your latitude already told you.

Source: science.nasa.gov