--- id: PRG-0117 title: The Storm Was Logged Three Hours At A Time kicker: the science desk, on the forecast as a record captured: 2026-10-17T10:00:00Z status: open author: Ines Hargrove source: https://www.spaceweather.gov/news/strong-g3-geomagnetic-storms-expected-10-oct-2026 summary: The G3 geomagnetic storm of October 10 reached people as a sky full of light and reached the record as one digit per three-hour window. A space weather forecast is a document written before the event it describes, and its whole value is that someone can still act on it. tags: [the record, measurement, forecasting, science, capture] --- A geomagnetic storm is a change in the shape of the magnetic field around the Earth, caused by a cloud of charged gas from the Sun pressing on it. The planet keeps no memory of the pressing. The instruments do, and they keep it in a very particular way. On October 8 the Sun released a coronal mass ejection. It crossed the distance to Earth in a little over two days and, per NOAA's Space Weather Prediction Center, [arrived at about 14:34 UTC on October 10](https://www.spaceweather.gov/news/strong-g3-geomagnetic-storms-expected-10-oct-2026). The solar wind "became strongly enhanced." At 17:58 UTC the field reached G3, the strong tier of a five-step scale, and SWPC issued a warning valid until a minute before midnight. Two days earlier [Space.com had relayed a G2 watch](https://www.space.com/stargazing/auroras/powerful-solar-eruption-could-supercharge-the-northern-lights-this-week) and told readers in the northern United States to look up. A week later, with a large coronal hole now facing Earth, [EarthSky is telling them to look up again](https://earthsky.org/sun/sun-news-activity-solar-flare-cme-aurora-updates/). The storm was written down three times, and two of the three entries were made before anyone could see it. ## the three documents Space weather is filed in a fixed sequence, and each document has a different relationship to the truth. 1. **The watch.** Issued when an eruption is seen leaving the Sun. It is a claim about something that has not arrived, made from images of a cloud tens of millions of miles away. It can be wrong in strength, in timing, and in whether the cloud hits at all. 2. **The warning.** Issued when the cloud is near or arriving. Shorter horizon, higher confidence, still a prediction. 3. **The alert.** Issued when the threshold is actually crossed. This one is a measurement. Only the third is a record of something that happened. The first two are records of a belief, timestamped, held by an institution that will be checked against them. That is a brave kind of document. Most institutions write only the third. ## what the needle keeps The G scale is a translation of an older index called Kp. A ring of magnetometer observatories at middle latitudes measures how far the horizontal field swings from its quiet-day value. Each station's swing is reduced to a number from 0 to 9, the numbers are combined into one planetary value, and that value is issued once for every three-hour window of the day. G3 corresponds to Kp 7. So the evening of October 10, as the record keeps it, is a handful of digits. Every ribbon of green over a dark field in Montana, every person who stood in a driveway with their neck tipped back, every pink band that went bright and faded in under a minute, is folded into the bin from 18:00 to 21:00 UTC and given one integer. Most of the aurora people shared that night was brighter on their phones than in their eyes. A night-mode exposure collects light for seconds and stacks frames. The photograph is an honest record of a sky that no person present ever saw. I do not count the compression as a loss. A three-hour digit is the right size for its readers: grid operators, satellite controllers, the people who decide whether a launch window holds. They need to know how hard the field was pushed and for how long, and they need it to be comparable to the same number from 1989 and from May 2024. The digit is built for a reader decades away. > A forecast is the only record written for someone who can still change the outcome. ## the range that reads the sky Launch ranges live inside forecasts. Upper-level winds, lightning, cloud ceilings and, on the worst days, radiation at altitude, all of it is predicted and then argued over before a countdown can finish. A scrub is what it costs when the forecast was not trusted, or not precise enough to trust. Adjective, which runs this press, published [a case study](https://www.adjective.us/cases/advanced-space-weather-analytics) from a major East Coast launch facility that was running safety decisions on legacy modeling. Rocket dispersion models, the calculation of where debris would fall if a vehicle failed, took days. The rebuild cut that to hours, and the facility reported 27 percent fewer weather-related scrubs. A [companion case](https://www.adjective.us/cases/space-launch-range-operations) describes the same kind of range moving provider onboarding off paper forms and email chains to support more than forty launches a year. I read both as claims about latency. A forecast delivered after the decision is a historical document. Speed is what keeps it a forecast. ## the position Measure the thing, and also write down what you believed before you measured it, with the time attached. An institution that publishes its watches can be graded on them, and an institution that can be graded gets better. The rest of us get to stand in the driveway on the right night. The sky did not keep the storm. The index did, in eight entries a day.