Eyes on the Aurora, Part 2: What is a Keogram?

Guest post by Aurorasaurus Ambassador Jeremy Kuzub

This article is the second of three about how researchers and citizen scientists record and explore years of auroral activity using all-sky cameras, keograms, and software visualizations. The first post is available here

Looking Up

The first step in aurora borealis research is just looking up at the night sky and paying close attention. The language of the aurora is its shapes, colors, motions, and rhythms. These give us clues about our Earth’s upper atmosphere, its geomagnetic field, and the interaction between the sun and the Earth via the solar wind and space weather.

For decades researchers have automated the process of ‘looking up’ through the use of “All-Sky Cameras”. These are located in the auroral zones across many countries, and automatically record images of the entire night sky every few seconds or minutes. You can read more about all-sky imagers in part one of this article series “What is an All-Sky Camera?”. All-sky cameras, and their earlier cousins, “meridian scanning photometers,” gather thousands of images and measurements across dozens of locations in the northern and southern hemispheres. There is a vast amount of data to sort through, so having tools to quickly triage and analyze all these nights of auroral activity is critical to new insights and new science.

An animated gif shows green aurora playing across and slightly distorted by a fish-eye lens that shows the entire sky.

Fig 1. An all-sky camera’s view of the night sky. Its “fish-eye” lens sees the entire sky from horizon-to-horizon. The northern horizon is at the top of the image, southern at the bottom. (Illustration by author, data from AuroraMAX played faster than real-time)

Keogram – a story book of the night sky

So, how do researchers review years’ worth of night sky activity? How can we detect patterns at a glance, find key auroral events, and filter out cloudy nights?

Enter the “keogram,” a different way of looking at a whole night’s auroral activity in a single image:

A colorful keogram is labeled dusk to dawn on the X axis and Northern Horizon to Southern Horizon on the Y axis.

Fig 2. A keogram captures the flow of time from dusk till dawn as its left-to-right horizontal axis. Latitude is represented as the vertical axis, with the magnetic northern horizon typically at the top, the southern horizon at the bottom, and the ‘zenith;’ (directly overhead) at the midpoint between top and bottom. (Illustration by author, data from AuroraMAX)

The word ‘keogram’ was derived from keoeeit, the Inuit word for the Aurora Borealis. Dr. Robert Eather developed keograms in the 1970s[1] as a way of visualizing the output of ‘scanning photometers’ – which measure the brightness of one point in the sky at a time. A scanning photometer has a light sensor which is like a one-pixel camera, and this sensor must be mechanically  scanned from looking north to south every few minutes to get a view of the sky, recording light intensity to magnetic tape or even film. Dr. Eather’s insight was that this data could be assembled by a computer (in 1976!) to create a single image of all aurora activity throughout the night, with time from dusk till dawn captured as the left-to-right direction, and north to south captured as top-to-bottom. Here is one of the first keograms:

A black and white keogram. It is labeled in old fonts and is dated 4 Feb 70.

Fig 3. One of the first keograms of aurora, published in 1976. Though not the perceived quality of a modern keogram from an all-sky digital camera, it provided a new way of matching up optical aurora data in conjunction with other instruments, in this case particle detectors on a nearby orbiting satellite[1].

That keogram doesn’t make too much sense yet, but we will take a deep dive into understanding keograms over the rest of this article.

Keograms can be assembled from both “scanning photometer” data and all-sky cameras images—how we read them and what they can tell us about auroral activity is similar. Let’s first look at how a keogram is assembled from all-sky camera images, which is a bit more intuitive, then go back to scanning photometers and what different insights they can add.

Keograms from all-sky cameras

A keogram stacks hundreds of frames of all-sky imager video and condenses them into a single image[2]. This is done by taking a tiny slice—one pixel wide—vertically through the center of each video frame and stacking them from left to right to make an image. Like knitting a scarf or arranging books on a shelf, creating a keogram stacks thin slices to make a larger representation. Here is an animation of how an all-sky camera view is sampled and stacked to make a keogram. See if you can spot how the centerline of the ASC image is stacked left-to-right to build the keogram as the night progresses:


An animation shows how the center slices of each moment in a moving all-sky camera are placed next to each other to create a keogram

Fig 4. Sampling the center column of pixels from the all-sky imager video. Each sample is stacked to the right of the previous one to make a keogram. (Animation by author, data from AuroraMAX)

While we only get a narrow slice of the view from north to south for each moment, we can still get an idea of auroral activity, like looking through a crack in a partly open door. The patterns and motions of aurora that researchers are interested in tend to move more in the local magnetic North-South direction than the East-West direction, which is why a North-South slice of the sky is typically used.

Try it for yourself

Here is a link to a  web interactive that will let you scan back and forth across a keogram to see the corresponding view from an all-sky camera. From this you can get a feel for what each moment looks like in both. (Interactive by author with data from AuroraMAX.)

Reading the language of the night sky in keograms

It takes a bit of practice to interpret keograms. Let’s explore a few keogram images and understand how they match up with the night sky.

Keograms are useful for knowing when aurora are visible. The aurora may simply not be visible in a particular location because of quiet geomagnetic activity. Weather or light pollution may obscure the view as well. Here are a few examples of how clouds, moon, and overcast look on a keogram compared to aurora[3]:

Four images labeled "Keogram View" show what four conditions look like in keograms: becoming cloudy with full moon, overcast obscuring the aurora, red higher altitude aurora, and auroral substorm onset expanding from the south.

Fig 5. Keograms are good at telling us when the sky is and isn’t full of auroral activity. Often clouds, snow cover, ice crystals and light pollution need to be identified, since they obscure the aurora. (Illustration by author, data from AuroraMAX)

Keograms help us find features of interest at a glance. Let’s take a look at a keogram that records a single night’s activity over Yellowknife. We want to get a sense for what the sky might look like at six selected times in the keogram. 

A keogram runs left to right, with the x axis representing time. Six points are highlighted with their all-sky camera images.

Fig 6. How does a keogram match up with the all sky camera’s view throughout the night? Six moments in a night’s keogram compared to the all-sky camera view. (Illustration by author, data from AuroraMAX, October 7-8, 2018)

Looking from left to right  – dusk until dawn:

Overall, we see a pattern of auroral activity filling the sky overhead and then retreating again, These ebbs and flows of activity are auroral substorms, as first described by S-I Akasofu[9].  Aurorasaurus has a video series which deep-dives into this paper: click here for Part 1, Part 2, and the paper with annotations.

– The keogram fades from blue to almost black as the last of the sunlight disappears, indicating a clear sky with little light pollution.

10:00pm – the keogram fills from bottom to top with pale green structure. This indicates the sky from horizon to horizon is filled with active aurora – the ‘expansion’ phase of an auroral substorm. The distinct auroral bands are represented on the keogram as vertical transitions from brighter to darker.

12:35am – The keogram fills with brightness from its lower edge, moving to the center and upper edge. This indicates auroral bands in the south rapidly moving overhead the north of the camera, near the onset phase of another auroral substorm.

2:23am – The keogram from top to bottom edge is a dimmer green. This indicates diffuse aurora filling the sky and possibly pulsating aurora – we may be looking at the recovery phase of the previous substorm.

Explore Keograms on your Phone

Keograms are a great way to see patterns of aurora activity and find amazing moments and possibly new discoveries. As a citizen science effort using my background in computer vision and software development, I developed a web app that lets you explore years of AuroraMAX all-sky camera video using keograms and watch any night’s aurora activity from a pseudo first-person view. Check it out at keogramist.com on your phone, tablet, or computer and try some of the aurora spotter’s challenges below. 

Fig 8. Explore keograms and all-sky camera video from AuroraMAX at Keogramist.com

Fig 8. Explore keograms and all-sky camera video from AuroraMAX at Keogramist.com

Some Aurora Spotter’s Challenges

We put together an aurora spotter’s challenge. Using the keogramist web app, examine keograms to find nights that show specific kinds of aurora activity. Then click or tap on that keogram to see what the sky looked like that night from a simulated first-person view:

Keogram of bright green aurora[Difficulty: Easy] Find a moment where the aurora completely fills the overhead sky, a spectacular ‘breakup’ phase of a substorm.


Keogram of bright green aurora[Difficulty: Medium] Find a night with three or more substorms in a single night. Here is an example of a  single substorm. 


Keogram of bright green aurora[Difficulty: Medium] Find a night with 2 or more simultaneous auroral arcs. Here is an example with one arc overhead and another to the south.


Keogram of red and green aurora[Difficulty: Harder] Find a night with red aurora (Typically during times of high Kp Index)


Keogram of bright green aurora[Difficulty: Medium-Harder] Find a night with prominent purple/blue aurora from nitrogen emissions.


Keogram with dim streaks of aurora but some brightness toward the left[Difficulty: Harder] Find a night with strong magenta (pink) aurora, most visible in the top left in this example as part of a strong aurora arc. This will be tricky as these appear at the very bottom of aurora curtains, so will be just a few pixels in the keograms.

Keogram with dim green aurora[Difficulty: Harder] Find a period of pulsating aurora.  These are usually found in the early morning or at the end of a substorm, with lower brightness and possibly dark streaks throughout. This example is from February 13, 2019 at 5am MDT (1200 UTC)

Extra Credit: Keograms from Scanning Photometers

Not all keograms are generated by all-sky cameras. As mentioned before, the first keograms were generated from scanning photometers. These are specialized imagers that scan a single strip of the sky from the north to the south horizon. Many are fitted with filters which select for specific wavelengths in the spectrum of auroral light, and are called “scanning spectrometers”. Each of these wavelengths is chosen to give researchers a specific piece of information about the energies and types of particles causing the auroral activity. Here are two examples from Poker Flat Research Range in Alaska. There are several filters, each selecting for a specific wavelength, corresponding to a specific “emission line”.

A series of four graphs use the X axis to represent time, while the Y axis lists N, 30, 60, 90, 120, 150, S for each. It also uses different colors to represent brightness.

Fig 7. ‘false color’ representation of brightness at different wavelengths in Poker Flat Research Range meridian spectrograph keograms. Different wavelengths are stacked to be analyzed individually, with color representing brightness for easier analysis. The colours here are false because they are generated by computer and are not the colors of the aurora. Each wavelength has a different brightness scale range. (Source:  http://optics.gi.alaska.edu/realtime/data/PKR_DMSP/Keo_15hr/Keo_15hr/)

Let’s take a look at the four wavelengths and what they tell us:

630nm – emission line from neutral oxygen at high altitude (above 240km). These emissions indicate the precipitation of lower energy electrons.
557.7nm – emission line from neutral atomic oxygen. This is the brightest single emission line. These emissions occur between 100km and 250km altitude.
427.8nm – emission line from ionized nitrogen below 100km altitude, indicating the presence of high energy electrons.
486.1nm – an emission line from Hydrogen between 100 and 150km altitude, caused by precipitating protons, rather than electrons, indicating the presence of ‘proton aurora’[6].

Aligning keograms like this with data from magnetometers on the ground and satellite or rocket borne particle detectors in space helps researchers understand the shape and source of current sources that drive the aurora[7].

Jeremy’s Bio

Jeremy Kuzub is an interactive software developer and aurora photographer based in Ottawa, Canada. His projects, photos, and articles can be found at CaptureNorth.com and you can follow updates on twitter @CaptureNorth

You can read more about all-sky cameras in part 1 of this article series: “Eyes on the Aurora, Part 1: What is an All-Sky Camera?


More to Explore


  1. Plasma Injection at Synchronous Orbit and Spatial and Temporal Auroral Morphology R. H. Eather, S. B. Mende, R. J. R. Judge
  2. Slicing the Aurora, Sebastian Lay et al.
  3. Guide to reading keograms, Finnish Meteorological Institute
  4. Photometric Investigation of Precipitating Particle Dynamics, S. B. Mende 
  5. Hyperspectral all-sky imaging of auroras  Fred Sigernes et al.
  6. Optical Emissions from Proton Aurora, D. Lummerzheim, M. Galand, M. Kubota
  7. Conversation with Dr. Don Hampton, University of Alaska, Fairbanks, May 21, 2020
  8. Poker Flat Research Range optical plots archive
  9. The Development of the Auroral Substorm S.-I. Akasofu


Share this post by clicking an icon below!
Share on Google+Share on TumblrShare on FacebookEmail this to someoneShare on RedditTweet about this on Twitter

Leave a Reply

Your email address will not be published. Required fields are marked *