## 3. THE CHECK THE BRIEF ASKED FOR: my cloud-for-cloud test (run 2026-10-31)

**Question.** Do frames from different nations at the same time show the same clouds?

**Method.** Downloaded published frames. For each, computed where every point on a 0.25-degree latitude/longitude grid falls on that satellite's disk, assuming a sphere and the satellite's stated position. Removed broad brightness with a local mean, so land and lighting drop out and cloud texture stays. Correlated pairs. Allowed a shift of up to 4 grid cells. **Control:** the same pairing with one frame from the day before, same hour, so lighting is alike and clouds differ.

| Pair | Time | Box | r same time | r control |
|---|---|---|---|---|
| NASA EPIC v Himawari-9 (Japan) | 2026-10-29, about 00:04 v 00:10 UTC | 120-180E, 35S-30N | **0.575** | 0.034 |
| Elektro-L No.4 file (Russia) v Himawari-9 | 2026-10-31, 03:00 UTC* | 130-175E, 35S-30N | **0.651** | 0.057 |
| Elektro-L No.3 file (Russia) v Meteosat 0 deg (Europe) | 2026-10-31, 15:00 UTC* | 50-18W, 35S-35N | **0.509** | not available** |
| Elektro-L No.3 file v GOES-19 (US) | 15:00 UTC* | same | **0.471** | 0.011 |
| GOES-19 v Meteosat 0 deg | 15:00 UTC | same | **0.426** | 0.016 |
| Himawari-9 v Meteosat IODC (45.5E) | 2026-10-31, 06:00 UTC | 90-118E, 30S-22N | **0.392** | 0.054 |
| FY-4B (China) v Himawari-9 band 13, infrared | 2026-10-31, 16:15 v 16:10 UTC | 100-150E, 30S-30N | **0.409** | 0.032 |

\* The Elektro site's file names run three hours ahead of UTC. I inferred Moscow time: file "1800" against GOES 15:00 gives 0.471; file "1500" gives 0.104. Inference from the fit, not a stated fact.
\** EUMETSAT's server would not return the day-before frame.

**What else it showed.**
- With a wrong sub-satellite longitude the correlation falls to about 0.01 (for FY-4B: 105E gives 0.41, while 86.5, 123.5 and 133E give 0.01). So the method finds the right place only when the geometry is right. The grid assumes a sphere seen from those positions, and the best shift was 0 or 1 cells. I read that as all these views fitting one sphere; I would not say it proves one, since a shared model would also fit.
- **Positions.** The agency's About page lists No.3 at 165.8E [raw] http://electro.ntsomz.ru/en/electro/about/ . RTL-SDR says No.3 moved from 76E to 14.5W in October 2026 [raw] https://www.rtl-sdr.com/elektro-l3-arrives-at-14-5w-and-begins-testing-l-band-lrit-and-hrit/ . The "l3" files fit 14.5W and the "l4" files fit 165.8E. The About page looks stale.

**Honest limits.**
- These are published products from agency servers: colour composites, not raw data, and not a signal I received. No home reception was done.
- Tuned by me against the match: EPIC rotation (best 0), GOES disk radius (906 of 903.5 half-width), FY-4B radius (1090). The controls used the same settings. Elektro and Himawari were not tuned.
- r is held down by viewing angle, parallax of cloud tops, band differences and JPEG. It is not a measure of how alike the clouds look; it should be read against the control.
- FY-4B compared at night, through a geocolour product against infrared, and only one frame ("latest" file).
- Not run: **India** (INSAT, not found openly), **Korea** (my GK-2A URL returned nothing; AWS bucket noaa-gk2a-pds is a US mirror), Fengyun in daylight, historic frames.
- One day's test, by me, unreviewed. A reviewer should re-run it.
- It answers "are the published pictures of one world?", not "who made them?". A theory could hold that all these agencies share one model of the sky and render it within minutes. What the check shows is the price of that theory: Russia, China, Japan, Europe and the US would all be in it, live, every 10 to 30 minutes.

**Rights warning.** Do not publish these frames until each agency's terms are read. NOAA and NASA are the easy ones; NICT, EUMETSAT, NTsOMZ and NSMC are not checked. The Elektro site also publishes an FTP archive with a login in its link; I did not use it.

**Public routes to repeat it.**
- Himawari: https://himawari8.nict.go.jp/img/D531106/latest.json and tiles under /img/D531106/2d/550/YYYY/MM/DD/hhmmss_x_y.png ; infrared /img/FULL_24h/B13/ ...
- EPIC: https://epic.gsfc.nasa.gov/api/natural
- GOES-19: https://cdn.star.nesdis.noaa.gov/GOES19/ABI/FD/GEOCOLOR/
- Meteosat: https://view.eumetsat.int/geoserver/wms (layers msg_fes:rgb_natural, msg_iodc:rgb_natural)
- Elektro: http://electro.ntsomz.ru/ (/i/splash_l3/, /i/splash_l4/, last 20 hours or so)
- FY-4B: http://img.nsmc.org.cn/CLOUDIMAGE/FY4B/AGRI/GCLR/FY4B_DISK_GCLR.JPG (latest only, time printed on the image)

---

