—valid time on the instruments
The Studio · nautical line · space-weather decision desk

Steer by the Sun.
Record the heading.

THE BINNACLE puts every coronal mass ejection in flight on the Sun–Earth line, shapes Earth's magnetic shield with the solar wind actually measured, keeps every forecast run as its own issuance, and seals each call with exactly what the instruments showed.

Who it is for

The person who has to make the call.

Five kinds of operation the Sun can force a decision on. The assets on the desk are a rehearsal of these calls, at real coordinates.

G G3
01 · GridUnder the oval

A corridor has to shed load when the auroral current arrives.

78°N south of 78°
02 · Polar flightHF goes dark

An R-scale blackout moves the track south of the polar cap.

6.6 Re Sun
03 · SatelliteNose inside GEO

When the shield crosses 6.6 Earth radii, the fleet goes safe.

RTK lost
04 · SurveyThe fix wanders

GNSS at high latitude fails the precision work, so the survey stops.

lost holdover
05 · TimingLeave the satellite

When GNSS scintillates, the order clock runs on its own oscillator.

And for whoever is asked, afterward, what was known when the decision was made.

Why I built it

So the heading can be proved.

The same ejection is forecast more than once, and the runs do not agree. The call still has to be made, and later it has to be shown. I built the Binnacle so that picture stays: each forecast kept as its own issuance, the magnetic shield shaped by the wind that was measured, and the decision sealed to exactly what the instruments showed. Official watches and warnings still come from NOAA.

01 · THE SUN–EARTH LINE

Every ejection, in flight.

Front aimed at EarthFront going elsewhereFlare on the disc, sized by class
02 · ISSUANCE

One ejection. Four forecasts.

The run you have selectedEarlier runs of the same ejection
03 · ARRIVAL

What actually came.

Shock pulse at Earth, from DONKI's interplanetary-shock log
04 · THE MAGNETOSPHERE

The shield gives way.

Magnetopause, ember when the nose is inside GEOAuroral oval from KpGeostationary orbit, 6.6 Re
05 · NOW

A quiet Sun, for the moment.

The data

Four public records. One instrument.

Public observations and published models are kept distinct. The ejections, runs, flares and shocks come from NASA's DONKI catalogue; the geomagnetic record from GFZ Potsdam; the solar wind and Dst from NASA's OMNI merge; the operational scales, the last week of L1 wind, the forecast and the auroral grid from NOAA SWPC, all taken . Where a gap is bridged by inference, the readout says so.

Ongoing monitoring

How the runs did against what reached L1.

Every WSA-Enlil run over the period that predicted an Earth arrival is scored against the shock DONKI logged for the same ejection: how many hours early or late, at what lead, and whether a predicted arrival came at all. This validation view describes the built-in data snapshot.

Arrival error by lead time

Each dot is one run: predicted arrival minus the observed shock, against how far ahead the run was made. The band is ± 12 h.

Sites and rules

Your assets. One rule each. In writing.

A transmission substation under the auroral oval, a polar air route on HF, a downlink in Svalbard, a survey vessel on RTK, an exchange colocation whose order timestamps hang on GNSS timing, a polar constellation. Each carries one operating rule on one metric and the action that follows, so no judgement is invented at three in the morning.

Three site groups ship with the desk, North Atlantic and Nordic, North America and the southern hemisphere, as synthetic assets at real locations. Replace them with yours; each is a few lines of JSON.

The record

Can you prove what the model said, why you acted, and who decided?

Every call is sealed into a SHA-256 hash chain with the state of both instruments at that moment: the valid time, Kp and its provenance, the measured wind and the magnetopause it gives, the ejection in focus with the run you had selected and what it predicted, the arrival outlook, every site's reading. Verify the chain on the spot. Export it as a studio.evidence.v1 pack.

1Genesis seals the rules and the data. The site thresholds and the snapshot's provenance are hashed before the first decision.
2Every decision carries its instruments. Not a link to them: the state itself, inside the block, including which run was on the glass.
3Each hash covers the one before. Change any block and every later hash breaks; the desk shows exactly where.
4The record survives the session. Stored with the page, verified on every load; tampering stops the seals. Export the pack, import it elsewhere, and the chain continues.
GENESIS#000 · 02:13:04Z6 sites · 428 CMEs · 352 runs4e244ab23cad…prev 00000000… SNAPSHOT#001 · 02:14:20Zvalid 04 Jul 05Z · Kp 7.0 · nose 5.8 Reb6334a09f6ed…prev 4e244ab2… DECISION#002 · 02:16:02ZReduce transfers · KRNKp 7.0 ≥ rule 7 · V 621 km/se0ade3bd5699… hash = SHA-256( prev + "\n" + canonical(body) )canonical = sorted keys · no whitespace · numbers to 6 dpverify → recompute every hash from genesis → ✓ chain verified
Under the hood

One file. Published physics. No backend to babysit.

The desk is a single HTML file with the 3D engine, the data snapshot and the evidence chain inside it. The built-in data snapshot supports replay. Direct public feeds can refresh where available; the original Vercel package also includes NASA and NOAA proxy functions.

earlier γ fitted 1 AU
Ejections

Each front follows the drag-based model of Vršnak et al. (2013), with the drag parameter γ fitted per run so the front reaches 1 AU exactly when that run's Enlil arrival says. Switch runs and the front moves.

bow nose Dst
Magnetosphere

Shue et al. (1998) magnetopause from the measured density, speed and Bz; a bow shock ahead of it; dipole field lines clipped where they cross the boundary; the ring current scaled by Dst.

day OVATION
Aurora

An analytic oval in dipole coordinates that widens and moves equatorward with Kp, dimmed on the day side, replaced by NOAA's OVATION grid at the hour it was issued.

±12 h lead drift
Verification

Every run with a predicted arrival against DONKI's shock log: absolute error, bias, share within ±12 h, hits, false alarms, misses, predicted against observed Kp, run-to-run drift.

010203 RUNSTATESEAL prev…prev…hash SHA-256( prev + canonical body ) studio.evidence.v1
Evidence

studio.evidence.v1 pack: {seq, type, ts, actor, payload, prev, hash}, SHA-256 over sorted-key JSON, verifiable in THE VERIFIER. The chain persists with the page, is re-verified on every load, and can be exported and resumed elsewhere.

index.htmlthis page app/the desk api/live feed NASA_API_KEY
Deploy

Drop the folder on Vercel. index.html is this page, app/ is the desk, api/ turns on the live feed; add a free NASA_API_KEY.

Open the desk. Scrub the storm. Seal a call.

Not a forecast productReal public data and published models, for decision rehearsal and record. The sites are synthetic. Official watches and warnings come from NOAA SWPC.THE BINNACLE · The Studio · nautical line