Research
Please can you help us?
To aid auroral researchers in the high arctic, the researchers from University College London run an all-sky camera in Kiruna, Sweden, that has taken an image of the sky through a fish-eye lens every 5 minutes for the last twenty five years. As you can imagine, that's a lot of pictures! For our research, we need to know when the skies are clear. Please can you help?
What's in an image?
The image above shows an 'all-sky' view of the sky above Kiruna. A 'fish-eye' lens mounted to the camera squeezes the whole sky into one circular view, with the horizon around the edge and overhead in the middle. North is to the top of these images. In this example image, you can see buildings and lights from the research facility, including one of the EISCAT radar receiver dishes (to the lower left). The sky looks fairly light but this is a long exposure. The clue to the darkness is that you can see stars in the background (if you know your constellations, you may even recognise a few!). The bright light to the right of the image is the Moon, and the wispy green band of light snaking across the image is the aurora (also known as the aurora borealis or Northern Lights). Around the horizon there are a few clouds but the sky in the middle (overhead) is clear. A satellite trail or meteor has also been captured in this image (the white line running down to the lower left of the image). Of course, not all images are this clear. In some the sky is cloudy, or the dome covering the camera is covered in snow.
What's so special about the Arctic?
The Earth's magnetic field provides a protective bubble around our planet, preventing energetic particles in space from stripping away our atmosphere.
The Sun, however, also has a magnetic field and this extends into space, carried by a constant outflow of particles from the the sun's atmosphere, known as the solar wind. Occasionally, vast eruptions of material, known as Coronal Mass Ejections (CMEs) erupt from the sun. CMES are fast (500-2500 km/s) clouds of hot, electrified gas (plasma) and magnetic field which, if they come towards Earth, can connect with the Earth's magnetic field, causing energetic particles to fall into the atmosphere along the magnetic field. The north and south poles are where these magnetic field lines emerge from the Earth (just like in a bar magnet) and so energetic particles are more easily able to enter the Earth's atmosphere here.
These particles excite the gasses in our upper atmosphere, causing them to glow (just as electricity causes the gas in a neon sign to glow) which we see as the aurora. Electrical currents flowing high in the atmosphere cause it to heat up and expand - making these 'solar storms' a hazard for spacecraft which are slowed down as they orbit through a temporarily thicker atmosphere. As currents flow at up to 2km/s, this accelerates the upper atmosphere, generating winds of hundreds of m/s (over 1000 miles per hour!) and temperatures over 1000 Celcius. The currents can cause electrical surges in power grids, while stirring up the atmosphere weakens the ionosphere (the electrified part of the Earth's upper atmosphere) causes problems for radio communications and GPS signals. These winds vary [with time of day, season and with solar activity] but in order that we can measure them, the skies have to be clear. This is because to measure the winds, we are looking at the faint glow given off by the upper atmosphere above 100 km known as 'air glow'. The airglow we are measuring is the result of oxygen atoms absorbing energy from sunlight during the day which it then releases as very specific wavelengths (colours) of light throughout the night. If the air is moving, these wavelengths are changed by the Doppler effect (in the same way that the siren of a police car changes pitch as it goes by - first moving towards then away from you). The airglow becomes redder if it is moving away and bluer if it is coming towards the observer. These tiny shifts are picked up by sensitive cameras and used to infer the speed of the winds. Airglow only occurs in the upper atmosphere however and to detect it, there must not be any clouds in the way, since these would scatter the light we are trying to detect, blurring out the Dopper shifts and degrading out measurements. (https://angeo.copernicus.org/articles/22/863/2004/). This is why we need your help to identify clear skies in order to make the best possible measurements of these winds.
Image courtesy of NASA.
Combining data from all the radar and cameras in the Arctic is helping us understand the impact of such 'space weather' on our atmosphere and protect these aspects of our modern technology on which we all rely.
How you can help us
The all-sky camera takes an image of the sky every few minutes throughout the year, whenever it is dark. We have processed these images so that the file sizes are small enough to share without breaking the internet but other than that, what you are seeing has not been processed. You will be shown sequences of three of these images in a short movie sequence. This helps to identify features such as clouds that move between images. Please tell us whether the the sky is clear, totally cloudy, partially cloudy or not visible.
Your classifications will allow us to be more confident in interpreting measurements of upper atmospheric winds and temperatures that, together with accompanying radars imaging the ionosphere (the ionised fraction of the upper atmosphere), will enable us to determine how efficiently energy from the solar wind is deposited in Earth's atmosphere during auroral 'space weather' events.
Is that it?
Well, there may even be other things you can see in the images such as the aurora, meteors, satellites or the moon. Tagging pictures with key words will enable us to carry out more detailed analysis of these images which could enable further science that is currently impossible!
Why not use artificial intelligence (AI) to do this?
Well, that would certainly be useful and indeed, through a few student projects we have attempted to use machine learning techniques to automatically identify images containing aurora and classify them by type. The results from such projects were mixed because auroral classification is somewhat subjective and identifying forms that were clearly distinct was not reliable. The current project is, in someways, more ambitious still, since we are trying to identify clear skies and yet computers find it challenging to distinguish between structured clouds and auroral whorls. We have come to the conclusion that to do this reliably, we need to train any AI by providing a set of images that have been classified by the most sophisticated computer we know of - you! We will start by providing you with hourly data in order that we can efficiently produce a clear sky index spanning the full timescale of the measurements (over 20 years). We will use these human-produced classifications to train a computer algorithm and test its reliability on an independent subset of the data. If this proves reliable we will apply the same technique to data from other all-sky cameras. If not, there will be many more images for Zooniverse volunteers to classify!
What does this have to do with Solar Stormwatch?
It seems a long time ago now that we launched the original Solar Stormwatch project in association with the Royal Observatory, Greenwich and the Zooniverse team at the University of Oxford. Indeed, it was 2010! In that project, volunteers classified images from NASA's STEREO mission in order to help forecast the arrival of CMEs at Earth. This work led to a successful PhD project and several publications in research journals. It has become a theme with Zooniverse projects that the insight and enthusiasm of volunteers often leads to unexpected results and Solar Stormwatch was no exception, with volunteers mapping the distribution of dust in space by identifying images in which interplanetary dust impacted the STEREO spacecraft.
The project was relaunched a few years later as Solar Stormwatch II in which the evolution in the shape of CMEs was tracked over time to see how they were being distorted by the solar wind. This led to a unique catalogue of CME evolution that is now being used to investigate how to improve solar wind forecast models such as HUXt, developed at the University of Reading. In conjunction with the National Science Museum in London, we also developed 'Protect Our Planet From Solar Storms' which invited volunteers to compare images of CMEs and tell us which looked 'more complex'. The result was another successful PhD and a research paper revealing that the visual complexity of such 'solar storms' followed the solar activity cycle.
Having enabled the science underpinning space weather forecasting (and dust!), we are now turning our attention to understanding the effects of these space weather events at Earth. Our experience tells us that, with the support of Zooniverse volunteers, we have the potential to answer science questions that would not otherwise have been possible. We do hope we can persuade you to join us in out latest project.