The norlys model is an experimental model that aims to transcribe global magnetometer data into a dynamic map displaying the probable, near-real-time auroral activity in the northern hemisphere (aurora borealis) and by extension in the southern hemisphere (aurora australis). This is the first-ever public aurora map of its kind featuring near real-time magnetometer data across such an extensive area. On top of assessing the near real-time activity, our norlys model features an archive of this activity over the last two hours.

The norlys model is derived from global magnetometer data, which are converted into a back-end ionospheric current map. This map works in the background and is updated every minute. It calculates the intensity and direction of equivalent ionospheric currents above an area of the world covering Europe, Iceland, Greenland and north America. The ionospheric current vectors are assessed thanks to magnetometer stations (provided by TGO, UAF, USGS, Carisma, IRF, FMI and DTU ; See Credits and acknowledgements), and derived from associated magnetic field displacements in the horizontal components (raw x and y), in nanoTeslas. This intermediary map enables us to locate the position of the real-time auroral oval, as well as to determine the flowing direction and intensity of the auroral electrojet current.

Some of these currents have been widely demonstrated to be closely assimilated to the auroral oval (specifically, the Auroral Electrojet current) and their direction / intensity correlated with atmospheric electron precipitation creating the aurora. Based on these scientific assumptions, our norlys model can give us a foundation to work out a relatively precise location and intensity for the auroral oval.

The norlys model gives a relatively accurate ‘nowcast’ of the location and intensity of the aurora activity and also enables users to go back in time to monitor the evolution of this activity over the last two hours. In a recent update, we have added two features to the model. [1] Full oval expansion where we extrapolate values longitudinally so that the oval appears more complete and [2] the southern hemisphere auroral oval, which is a only a mirrored image of the northern hemisphere oval thanks to their conjugacy. Please read the disclaimers (below) carefully, as these two functions have pros and cons!

NB: Our model favors the stronger auroras to the detriment of very weak ones, hence the scale displaying the probability of strong auroras. Weak aurora activity may still naturally arise along the oval and the polar cap whereas magnetometers (and so ultimately our model) don’t display much or anything at all (please read disclaimers carefully).

By computing ionospheric current data, we are able to output a naturally realistic map of the tentative auroral oval, which evolves in latitude, longitude and intensity in near real-time. This is extremely useful for aurora chasing because users can literally locate where the oval is and where the maximum activity can be found along the oval right then and there. This is especially true when users monitor the evolution of the oval on our map over several minutes, as discrepancies may arise at times (see disclaimer below). With the ability to follow the auroral activity over the past two hours, users can easily identify how far from the oval they are, assess the sector of the oval they are crossing, understand the evolution of the auroral activity and eventually anticipate the type of auroral activity that will occur near them.

Please note that our model does not precisely determine the different types of auroral behaviors along the oval (e.g. quiet arcs, substorms, pulsating auroras…) because it is not (yet) possible to use magnetometer / current data to do so in a very scientifically accurate way. If it was, such models would likely have already been produced. With that being said, our model can still help greatly to anticipate the major types of behavior using the relative intensity and location of this activity.

As an example, early magnetic evening auroras are often quiet arcs and so the geographical locations passing under this sector of the auroral oval should on average only experience a thinner, dimmer activity in our model. In the same way, closer to magnetic midnight, the nightside of the Earth experiences more explosive auroras (substorms) that work in a dim build-up (growth), then bright explosion (expansion), followed by a diffuse phase (recovery). Our model does show the real-time development of these spectacular substorms by sudden changes in intensity and location of the oval. The growth phase will display a low intensity aurora slowly migrating equatorward, typically over several hours. The substorm oval will likely remain dim and thin just like during early evening. The expansion phase will typically feature a sudden, quick (minutes), intense brightening (highest values) of the aurora traveling from the equatorward oval towards a much more poleward location. The recovery phase will generally show a moderate intensity but over a much larger oval in latitude span.

The model has been designed with the utmost care for user utilization. It is pasted onto a 3-D globe that can be smoothly rotated, zoomed and moved. Use your mouse (desktop) or fingers (mobile) to drag and rotate the globe. Scroll forward (desktop) or pinch out (mobile) to zoom in. Scroll back or pinch in to zoom out. As you load the homepage, the model should center itself on your current location automatically, provided you allowed location service when you first downloaded the app / used the website. After a certain time of inactivity on the page, our model goes into a ‘showcase mode’ where different webcams around the world are displayed. This mode deactivates as soon as you move your mouse and use the map.

The map also shows cities where we have online webcams, as well as your location (provided you turn on location services). On the desktop version, hover over the city to reveal the webcam. On mobile devices, click on the city to reveal the webcam. Click again to close it down.

As the sunlight is the major obstacle that prevents one from seeing the aurora, our model displays where the sunset / sunrise terminator is. It is the line that marks the transition between the lightest gray area of the map and a band of darker gray. This band of darker gray corresponds to a time after sunset or before sunrise where it is yet too bright for aurora (the Sun is between 0 and 8 degrees of elevation under the horizon). The darkest color (very dark gray) corresponds to the area of the Earth where it should be dark enough to see the aurora. In this zone, the Sun is below 8 degrees of elevation under the horizon. Please note that from 8 to 18 degrees under the horizon, it is not fully dark (twilight), Consequently, it can affect aurora viewing by washing out the fainter auroras.

It is important to note that our model is calculated using magnetometer station inside the area delimited by a dashed gray line in the northern hemisphere. Any auroral activity outside this zone comes from strong interpolation, so by definition not very accurate.

The scale of aurora probability goes from low to high. Very low to moderate activity would correspond to faint aurora, often encountered when the geomagnetic activity is generally quiet, or during substorm growth. This type of aurora is often seen in the daytime oval (dayside aurora), in the afternoon oval (discrete quiet arcs), the early evening oval (discrete quiet arcs), the late evening and nighttime oval (substorm growth arcs or substorm recovery pulsating aurora). Moderate probability often describes situations like more intense evening auroras, more intense substorm growth and substorm recovery. High and very high intensities typically correspond to the most active auroras, occurring generally during substorm expansion in the nighttime oval, during other types of explosive auroral activity (convection bay), at the beginning of the recovery phase or anywhere along the oval under a unusually active magnetosphere ( called geomagnetic storming, where all the different sectors of the oval intensify).

Users can click on any point of the viewing grid to reveal the aurora activity score (NB: not available in the southern hemisphere for now).

Use the slider to reveal the state and location of the aurora activity at any time during the last two hours. You can also let it play as a timelapse movie by pressing on ‘play’ on the left-hand side of the slider. Using the slider may introduce a little lag in displaying the aurora data, as there are more than 200 data points to retrieve per minute. Please allow some time for data retrieval before playing. NB: the ‘play’ option will only display rougher boundaries for countries and aurora activity because of the amount of data. You can come back to the nowcast by clicking on ‘latest’ on the right-hand side of the slider.

Our model is not a forecast. While you can use the model by spotting where the current aurora activity is at a glance, we recommend that you monitor it closely over several minutes (or dozens of minutes). As magnetometer data can display very quick changes or errors, our model may therefore underestimate or overestimate the auroral activity, resulting in gaps, discrepancies or sudden drastic changes. These are almost always erased naturally in the course of several minutes, hence the need to monitor the map on a regular basis and not just punctually. The time bar makes it much easier to study how the auroral oval evolves over two hours, enabling you to better anticipate what is likely to occur next.

We are eternally indebted to Dr. M. G. Johnsen, head of the Tromsø Geophysical Observatory (TGO) at UiT The Arctic University of Norway, along with his team for providing us with geomagnetic data from Norway, Greenland and Svalbard, as well as helping in the construction of several aspects of the model. We would also like to thank the Finnish Meteorological Institute for providing geomagnetic data in Finland. In the same way, we thank the IRF (Swedish Institute of Space Physics) for the Swedish data, and the DTU Space (Danish Institute of Space Physics) for the Danish data. Finally, we want to extend our special thanks to the UAF (Alaska, Fairbanks), the USGS (USA) and Carisma (Canada) for providing us with the North American data.

[1] It’s a citizen-science initiative: the present model is merely experimental. It means that although it bases its theory and data on science literature and scientific observations, the model itself has not been peer-reviewed and is not associated with any scientific / governmental institution.

[2] Very good but never 100% accurate: strong ionospheric currents like the Auroral Electroject and the magnetic field variations they create are often associated with auroral activity. However, there is still a wide gap between ionospheric currents data and real-life auroral activity. When current-generated field variations are intense, the model becomes more accurate but quiet times may introduce some bias. Thus, although our model has been tested to present a thorough estimate of the auroral oval and its activity, there may be some inaccuracy and discrepancies in the intensity and location of the auroral activity at times.

[3] To follow up on the last point, you may experience some auroral probability ‘blobs’, which are far from the oval. Please note that although we have implemented a code that limits errors, the raw magnetometer data are provided by different institutions around the world and there are often significant differences in the instrumentation used. Therefore, although we recalibrate all data following an advanced baseline calculation, there may still be data errors and discrepancies from time to time, showing you zones of aurora probability that seem out of place. They are easily identified by their ‘blobby’ look and by their distant location to the auroral oval.

[4] Delays: the model is updated every minute by retrieving magnetometer data. The latter already presents a form of delay compared to the real-time local magnetic activity. In the field, most auroral activity is evolving slowly, over dozens of minutes, if not hours. This makes the use of magnetograms rather easy and timely. However, you may regularly experience brightenings of the aurora (substorm expansion for example), which only last for seconds or minutes. These quick intensifications tend to appear with a delay in a magnetogram array and might consequently be finished before they can be read in our model. However, there is a double silver lining to this. A) A lot of these brightenings are still slow to develop (example during solar storms), enabling our model to catch up and show you an increase in auroral brightness in time. B) By following our model and the time bar regularly for the nightside oval where these intensifications happen frequently, users can put themselves in the best position to catch the brightenings by identifying the build-up process (growth), which is characterized by a thinner, dimmer oval slowly migrating equatorward.

[5] Spatial Interpolation within our calculation area: The oval representation may be spatially inconsistent at times because some countries / land masses are simply not covered by magnetometer data or we haven’t been able to gain access to magnetometer data for such areas. Therefore, a stronger interpolation had to be implemented in such zones, leading to more spatial inaccuracy. Example: we do not yet have access to Icelandic data and so there is a large interpolation going on between Europe and Greenland, making the longitudinal resolution less accurate for iceland. In the same way, we have much fewer magnetometers at lower latitudes, so the representation of the oval will become much more uncertain when the oval really expands equatorward during extreme geomagnetic storms. In such cases, it shouldn't be too much of an issue as other features of our app would notify you of the very intense geomagnetic conditions at mid-latitudes.

[6] Strong spatial Interpolation outside of our calculation area: In a new update, we have implemented an algorithm that extrapolates strongly between magnetometer-rich and magnetometer-poor zones, so that the oval appears more complete. This is especially the case in the zone outside our calculation area in dashed gray line, i.e. mostly Russia. Please note that interpolated values are much less accurate in time, in place and in intensity in these zones.

[7] Southern hemisphere oval: It has long been demonstrated that the northern auroral oval and its southern counterpart are mirrored. We call this property auroral oval conjugacy. In Layman’s terms, it means that the aurora borealis and the aurora australis should theoretically produce the same activity at the same time and at the same place. Many of our early users wished to get access to the southern hemisphere oval and this very property enables us to do just that despite the lack of regular magnetometer stations to draw the southern oval correctly. We pasted and fitted a mirrored image of the northern oval calculated by our model in the southern hemisphere centered around the southern geomagnetic pole. Users in the southern hemisphere can now have an estimation of the oval according to their location. However, a word of caution. Despite oval conjugacy, many recent studies have also pointed out differences arising in between the two ovals, some of which are thought to come from solar wind properties, the magnetic configuration of the IMF (By, Bx), magnetospheric properties and even sunlight (conductivity in the ionosphere)! Therefore, we advise our southern hemisphere users to take this representation of the southern oval with a grain of salt. We recommend that you roughly follow the oval in its entirety, its global intensity and its latitude, and that you give less attention to longitudinal accuracy. Following this simple rule should limit the number of errors. We are still working to implement magnetometer and webcam data into this model to render it more accurate in the future.

[8] Lack of data: some users get surprised or frustrated to see very weak or no auroral activity on our model on a regular basis. But that’s how the aurora works! Most of the time, the aurora oval is quiet and gets enhanced only on sporadic occasions (except during geomagnetic storms). Our model is not a forecast like the OVATION, which shows some continuous auroral activity. Our model is a nowcast and an archive of the real-life auroral activity, so it is normal that it appears weak or empty a lot of the time. Although it is almost impossible for the oval never to contain any auroral activity, there are many times where the auroral activity along the oval is very weak or almost absent. In such cases, magnetometers will not feature many variations and/or are close to their baseline. Therefore, our model will not show much activity or anything. Such instances occur often under generally quiet solar wind conditions, and/or in the evening oval, and/or in the polar cap, and/or during the growth phase of weaker substorms. In such situations, auroras may still occur in the oval in weaker forms whereas our model does not display much visually. That’s why our intensity scale is based on strong auroral activity. When you don’t see anything on the model, it is important that you A) click around on the map to detect very weak values that are hard to detect visually (typically values between 0 and 1, showing a very weak green), B) monitor the model continuously over several minutes to spot the reappearance of the oval, C) check webcams regularly to verify the evolution of the oval when the model fails to show it. As soon as the magnetometer activity gets stronger again, the oval should reappear. Working with magnetometers has its advantages but it also has its drawbacks, and this is one of them. We are only interpreting what magnetometers register, nothing else!