Geomagnetic indices explained, side by side

Six different numbers get called “the aurora index” in casual use, and they measure different physical quantities, on different clocks, from different observatories. Choosing the wrong one is why a forecast can look excellent while the sky does nothing, and why an archive value can contradict what an app showed at the time.

A geomagnetic index is a number summarising how much Earth’s magnetic field has been disturbed, computed from magnetometer readings on the ground or, in one case, defined as a severity scale on top of another index. None of them measure aurora. They measure magnetic disturbance, which is a consequence of the same process that produces aurora, and the distance between those two things is where most confusion lives.

What each index actually measures

The table below sets out the six numbers that appear in aurora contexts. Read the cadence column first: it explains most of the disagreements between sources.

Kp: the three-hour planetary summary

Kp is a planetary index of geomagnetic disturbance reported for each three-hour UT interval, on a quasi-logarithmic scale from 0 to 9 subdivided into thirds. NOAA’s National Centers for Environmental Information describe it precisely: the mean standardised K-index from 13 geomagnetic observatories sited between 44 and 60 degrees northern or southern geomagnetic latitude.

That siting is a design decision, not an accident of where instruments happened to exist. The stations sit deliberately below the auroral zone, so the index reflects broad planetary conditions rather than the violent local swings a magnetometer inside the oval records. It is the reason Kp is a good planetary summary and a poor local aurora detector.

The scale is expressed in thirds: 5− is 4⅔, 5 is 5, and 5+ is 5⅓. The single-station K index it is built from was introduced by Julius Bartels in 1938, and the Kp and ap series published by GFZ extends back to 1932.

The three-hour window is the property that matters most and gets discussed least. Eight Kp values cover a UT day, running 00–03, 03–06 and so on. An auroral display that is spectacular for 25 minutes and quiet for the rest of the block produces a modest Kp, because the index is describing the block, not the 25 minutes.

That is not a defect. Kp exists to give geophysicists a consistent long-baseline record of planetary disturbance, and it does that better than anything else because the method has stayed stable for over eight decades. It was never designed to tell one person whether to put a coat on. The Kp index explained, and what it cannot tell you covers that mismatch at length.

ap and Ap: the same thing, linearly

The ap index is the linear equivalent of Kp for the same three-hour interval, expressed in nanotesla. Ap, capitalised, is the arithmetic mean of the eight ap values for one UT day.

The reason both exist is arithmetic. Kp is quasi-logarithmic, so the interval between Kp 7 and Kp 8 represents a far larger change in the underlying magnetic disturbance than the interval between Kp 1 and Kp 2. Averaging Kp values produces a number with no physical meaning. If you want a daily figure, or a monthly one, ap and Ap are the correct inputs and Kp is not.

This distinction is routinely broken in popular write-ups that quote a “daily average Kp”. If you see one, the underlying calculation is either wrong or was actually done on ap and relabelled.

Hp30 and Hp60: Kp at the timescale aurora runs on

Hp30 and Hp60 are high-cadence geomagnetic indices published by GFZ on a scale comparable to Kp, computed over 30-minute and 60-minute intervals. They are not capped at 9.

Both properties address real limitations. The 30-minute window is much closer to the timescale on which auroral activity genuinely varies, so a substorm that Kp averages away appears as a distinct peak. The uncapped top end means the most extreme storms are not all flattened into a single value of 9, so events of very different severity remain distinguishable in the record.

For an observer, Hp30 is the closer analogue of “was there something happening in the last half hour”. For a historian of the record, Kp remains the index with the long baseline, and that is why both are published rather than one replacing the other.

Dst: how deep the storm is

Dst, the disturbance storm time index, is an hourly measure in nanotesla of the depression in Earth’s horizontal magnetic field caused by the ring current. It is computed from four near-equatorial observatories and maintained by the World Data Center for Geomagnetism in Kyoto.

Dst answers a different question from Kp. It describes the strength of the ring current — energetic ions drifting around Earth at a few Earth radii — which is the standard measure of a magnetic storm’s depth. Storm classifications in the research literature are usually stated in Dst, not Kp.

The timing catches people out. Dst reaches its most negative value at the end of the storm main phase and then recovers slowly over a day or more. Auroral activity is frequently at its most vigorous during the main phase, before the minimum, and the long recovery tail can show a strongly negative Dst on a night when the sky has gone quiet.

AE: the index that moves when the aurora moves

AE, the auroral electrojet index, is computed at one-minute cadence from a chain of observatories inside the auroral zone. It tracks the strength of the electrojet current flowing in the auroral ionosphere at about 100 kilometres altitude.

Of the standard indices, AE is the one that responds on substorm timescales, because it is measuring the current system that a substorm switches on. Its limitation is coverage: the station chain is fixed, so the index sees activity well when it happens over the instrumented longitudes and less well when it does not.

The NOAA G-scale: severity with consequences attached

The NOAA G-scale is not an independent measurement. It is a five-level severity scale defined on top of Kp, and its value is that each level carries a published statement of documented effects and an average frequency.

Two things in that table are worth reading carefully. The first is that the latitudes quoted are geomagnetic, not geographic — which is exactly the distinction that makes city-to-city comparison unreliable, covered in why your neighbour sees more aurora.

The second is the frequency column. G1 conditions occur on roughly 900 days of an 11-year cycle, which is a quarter of it. G5 occurs on about four days per cycle. The gap between “aurora is possible” and “aurora is possible where most people live” is nearly three orders of magnitude in frequency, and no amount of app design closes it.

Why two sources can show different Kp values for the same hour

NOAA’s Space Weather Prediction Center publishes an estimated planetary K-index in near-real time, because operational users need a number now rather than a definitive one later. GFZ later publishes the definitive Kp for the same interval, computed from the full set of observatory data after quality control.

The two can differ. That is not an error in either — one is a real-time estimate and one is a final value, and the difference is usually small. It does mean that an app showing “Kp 6 right now” and an archive later showing Kp 5+ for that interval are both reporting correctly.

A second source of disagreement is that some services display a station K value rather than the planetary index. A K value from a single observatory describes the disturbance at that observatory, which can be considerably larger or smaller than the planetary summary. If a number looks surprising, check whether it is planetary or local before concluding anything.

The one with the longest memory: aa

There is a seventh index worth knowing about even though no aurora app displays it. The aa index is a three-hourly equivalent amplitude computed in France from just two nearly antipodal observatories, one in England and one in Australia, and NOAA’s archive covers it from 1868 to 2010.

Two stations is a much weaker basis than the 13 behind Kp, and aa is correspondingly noisier. Its value is the baseline: it reaches back more than six decades further than Kp does, which makes it the index used when someone wants to compare a modern storm against the nineteenth century.

Which index answers which question

Match the index to the decision rather than looking for the best one.

• “Is it worth going outside in the next hour?” None of these. Use the interplanetary magnetic field Bz measured at L1, which is described in how to read solar wind data, and check the sky. Indices describe what already happened.
• “Did something happen in the last half hour?” Hp30, or AE if your longitude is covered by the station chain.
• “Was last night active?” Kp, read as a sequence of blocks rather than a single number.
• “How big was this storm compared with historical ones?” Dst for depth, and the G-scale for a severity label with consequences attached.
• “How active has this month been?” Ap, never averaged Kp.
• “Could I have seen it from where I live?” The G-scale table above, read against your corrected geomagnetic latitude rather than your map latitude. The geomagnetic latitude calculator gives you that coordinate.

The limitation all six share

Every index on this page is a ground-based measurement of magnetic disturbance, and magnetic disturbance is a proxy for aurora rather than a measure of it. The chain from one to the other has several links that indices do not see.

An index does not know where you are relative to the active sector, because a storm can be concentrated over one range of longitudes. It does not know whether it is dark where you are, which above the Arctic Circle rules out months at a time regardless of activity — see why the aurora has a season. It does not know whether the display is bright enough for human vision rather than only for a camera. And no index has ever known anything about cloud, which ends more aurora nights than every other factor combined.

That is the reason this site’s tools score locations rather than displaying a number: an index is an input, not an answer. Can I see the aurora tonight combines activity with your latitude and darkness rather than reporting Kp and leaving the rest to you.

Bottom line

Kp is a three-hour planetary summary on a quasi-logarithmic 0–9 scale from 13 subauroral observatories, published by GFZ in Potsdam; it is the long-baseline standard and it averages away exactly the substorm timescale on which aurora varies. The ap index is its linear equivalent and the only correct input for averaging; Ap is the daily mean of ap. Hp30 and Hp60 apply the same approach at 30 and 60 minutes without a ceiling at 9, which makes them better suited to observing. Dst measures ring-current depth hourly from four near-equatorial stations and reaches its minimum after auroral activity often peaks. AE tracks the auroral electrojet at one-minute cadence and is the fastest-responding standard index. The NOAA G-scale maps Kp onto five severity levels — G1 is Kp 5 at about 1700 events per cycle, G5 is Kp 9 at about four — each with published effects and lowest reported latitudes.

For deciding tonight, none of them beats a measured southward Bz plus a clear sky. For understanding what happened, pick the index whose cadence matches the question.