The first round of Sunspotter data classification has been completed!
It took around 1,639 registered volunteers (+ an unknown number of anonymous volunteers) 324,465 clicks to compare each image in the set of 12,966 data at least 50 times. So thats an average of 198 clicks/registered volunteer!
Of course in reality, there are those few Sunspotter-aholics that probably did a lion’s share of the work- To everyone that wants to get going on comparing the next set of images, we are working as fast as possible to get the next set of data up!
The forthcoming dataset relies on sunspot groups detected by a completely automated algorithm (not relying on human intervention, like the last dataset), called the SolarMonitor Active Region Tracker (SMART; Higgins et al. 2011), and will not be prone to human bias. It will include over two hundred thousand images of sunspot groups. A number of the sunspot groups will overlap with the last dataset, but will be detected and processed in a different way.
As mentioned in a previous post we are using ‘Stereographic Projection‘ techniques to ‘de-smoosh’ the sunspot groups near the solar limb (edge), so that they appear as if they were at the center of the Sun. Stereographic projections are rarely used for images of the Sun (if anyone knows of an example, please tell me!), but they are common in astrophysics). Also, they are commonly used to make maps of the South and North pole of the Earth because although features at the edge of the map become larger by a factor of ~2, they keep shapes the same (the more usual Mercator maps completely distort shape near the poles). Keeping shape the same, or being conformal, is important for Sunspotter, because the shape of a sunspot group is likely to have a big effect on apparent complexity!
In addition to being the first time that anyone has measured the ‘true’ complexity of a large sample sunspot groups, we will now be able to measure the evolution ofsunspot group complexity, and determine how that relates to eruptions.
In the mean time, we will be analysing the previous dataset to determine how complexity relates to other properties of sunspot groups and to solar flare occurrence. Exciting times!
A big thank you to all of our volunteers who helped us to complete this awesome dataset!
Magnetograms are wondrous things, that when I stop to think about what they are, I can scarcely believe that they even exist! A magnetogram is an image, where color represents magnetic field strength. The idea that you can take a picture of a magnetic field completely goes against my intuition of how magnetic fields work. Prior to studying physics, I only knew magnetic fields as funny forces one feels when pushing/pulling two refrigerator magnets together/apart. How could one possible take a picture of an invisible force!? Well, I’m not going to lie- its not a simple process at all. But I’ll give the explanation my best shot..
I will space out this explanation over a few posts, so I can go into a bit of detail about the whole process- starting from the ‘basic’ physics concepts. …I’ve never met a physics that I thought was basic 🙂
While reading this, if you have any corrections (I’ve never read or written a physics discussion with out at least one mistake) or questions (no matter how ‘basic’ you think they are), please post them in the comments section or on Sunspotter Talk!
Magnetic fields put the ‘magnet’ in Magneto
I could try and describe magnetic fields in some arcane abstract way, but I will start with something everyone has at least heard of: light… A magnetic field is one component of a light wave; the other component is an electric field. If you remember nothing else from what follows, remember these three rules:
- a changing electric field generates a magnetic field; (formally Ampere’s Law)
…coiling an electrified wire around a magnet will cause it to spin = an electric motor
- a changing magnetic field generates an electric field; (formally Faraday’s Law)
…passing a magnet through a coil of wire will generate electricity = an electrical generator
- an accelerating (or decelerating) electric field generates light. (e.g., synchrotron and Bremsstrahlung radiation)
And what is light? Incredibly, light is an oscillating electric field and an oscillating magnetic field that perpetuate each other, as long as they travel at the speed of light, 300,000 km/s. This is a constant conversion of electric energy into magnetic energy and magnetic energy back into electric energy. The energy of light is determined by how fast the electric and magnetic field is oscillating. However, in quantum mechanics, light is represented as photons, which are light-wave ‘packets’ of energy, that behave like particles (e.g., they can bounce off of another particle, like an electron). But in classical mechanics, light behaves as a wave (e.g., two light waves can ‘interfere’ and cancel out or build in strength, just like water waves). The fact that light can simultaneously behave as both a particle and wave is known as ‘wave-particle duality‘. To make a magnetogram, you sometimes have to consider light as a particle, and sometimes as a wave.
Like a packet of crisps (chips), a photon packet of light is VERY calorie dense. Each photon has a specific amount of energy stored inside. If the conditions are right, a photon can deliver all of its energy to an atom (e.g., a hydrogen atom) by smashing into it. When it does this, the orbit of the electron going around the proton (in a hydrogen atom, for example) will become more energetic, and the photon will be completely absorbed by the atom. This is called photon absorption. On the other hand, an energetic electron in an atom can suddenly (sometimes randomly) become less energetic, and in doing so, release a photon. This is called photon emission, and releases photons at a very narrow range of energies that correspond to the change in energy of the electron’s orbit.
There is another very different way that light can be released. When anything glows (like a red-hot iron or an old fashioned light bulb) a lot of photons are being released, but in a different way than with ‘emission‘. When something glows because it is hot (like the surface of the Sun), it is usually releasing thermal radiation. When matter has a high temperature, it means that the particles it is made of are moving around really fast, and banging into each other. Remember Rule #3? Every time charged particles in the matter bang into one another, they undergoing deceleration or acceleration, like the balls bouncing around on a pool table, and they release light! Not surprisingly, the hotter the matter is, the more thermal radiation is being released. It turns out that when something glows at a certain temperature it will release a predictable fraction of its photons at each energy; we can use Planck’s Law to predict this. Thermal radiation releases photons at many different energies (a broad spectrum).
The main thing to remember is this: photon emission releases light in a narrow range of energies, while thermal radiation releases light waves in a broad range of energies.
Proton says to a Neutron, ‘I’ve lost my Electron!‘
The Neutron says, ‘Are you sure?‘
The Proton replies, ‘I’m positive!‘
The surface of the Sun is like an ocean, but instead of being made of water molecules, it is made of (mostly) hydrogen and some helium. Another difference between the solar surface and an ocean is that instead of a refreshing ~290 Kelvin the surface of the Sun is a blazing 5,700 Kelvin (…so hot that it glows yellow!). For this reason, it is a plasma (and not just a ‘gas’). It is a plasma because a significant fraction of the Sun’s hydrogen has been stripped of its electron, leaving just a lonely ionised proton (a positively charged particle), wandering around looking for an electron. This gives plasma some very interesting properties, which will be covered in a future blog post when we discuss the physics of sunspots.
The Sun isn’t just hydrogen, it also has a small amount of a lot of other elements. These other elements can also undergo photon emission, which produces photons of certain energies. We observe this light as emission lines. By studying emission lines, Helium was first discovered because we saw one of its lines when we looked at the light from the Sun (the name Helium comes from Helios, the Greek Sun god). Atoms or partially ionised atoms (that still have at least one electron) undergo photon emission and absorption all the time because of all the other particles and photons smashing into them.
When atoms or ions undergo photon emission in the presence of a magnetic field, something strange happens. And because of this we are able to measure magnetic fields on the Sun- but you’ll have to wait for The Making of a Magnetogram Part II to hear all about it!
As always if you have any questions/ comments /suggestions, we are eager to hear them!
Thanks for listening.
I am a space weather research scientist at the Met Office*, the national meteorological service for the United Kingdom. My job is to transition basic research to operational forecasting- in other words, I try to improve space-weather forecasts with new-and-improved science!
What do I mean by space weather? It basically describes the changing environmental conditions in near-Earth space. Magnetic fields, radiation, particles, and matter, which have been ejected from the Sun, can interact with the Earth’s upper atmosphere and surrounding magnetic field to produce a variety of effects.
There are streams of particles from the Sun constantly hitting the Earth via the solar wind, but the Earth experiences an increased impact during periods of high solar activity, when solar eruptions can occur in the form of solar flares and coronal mass ejections (CMEs). Solar flares are sudden releases of energy across the entire electromagnetic spectrum. They are hard to predict, and the energy can be detected in the Earth’s atmosphere as soon as 8.5 minutes after a solar flare (travelling at the speed of light). CMEs are often associated with flares, eruptions of large amounts of matter from the solar atmosphere. These can take days to reach Earth, carrying a local magnetic field from the Sun. Considering the short time frame for forecasting of flares compared to CMEs, it’s really important to have accurate alerts for big events. That’s where your help with Sunspotter comes in – by improving our understanding of the active regions that are the source of flares, we can hopefully improve our forecasts!But why do we care about space weather? In our increasingly technologically-dependent society, the impact of solar eruptive events can actually be quite severe. Some key sectors in need of accurate event forecasts include energy, aviation, satellite operation, marine, communications, rail, and defence. For example, interruptions to radio communications and GPS can occur due to eruptions, and power grids can also be disrupted. Particles accelerated during eruptions can also damage spacecraft and degrade electronics, and instruments often have to be switched-off or reset. It is important for satellite companies to receive accurate information on the likelihood of eruptions to ensure as little down-time as possible. In the aviation industry, flight crews, passengers, and onboard electronics are all under direct exposure to higher levels of radiation on transpolar flights. Even astronauts cannot take space walks during these events. Solar-activity monitoring systems are imperative to keep astronauts safe! There have been a number of events directly related to high solar activity in recent years. For example, during a particularly large event in 1989, the entire power grid of Quebec collapsed, causing a 9-hour blackout which effected 6million people. In December 2006, a powerful flare disrupted satellite-to-ground communications and GPS navigation system signals for about 10 minutes. The eruption was so powerful it actually damaged the solar X-ray imager instrument on the GOES 13 satellite that was taking images of it! An American telecommunications satellite, Galaxy 15, now widely known as ‘zombiesat’, ceased responding to commands in 2010. The manufacturer has theorised that solar activity was responsible for the satellite malfunctioning, although they could not settle on a single root cause. Check out the National Research Council or the Royal Academy of Engineering reports for more information on, and examples of, space weather impacts.
In order to monitor and forecast space weather, we generally use ground-based and satellite instrumentation. The solar surface and atmosphere is observed in near-real time to detect any new active regions that may become the source of large events. These observations, such as the MDI images used in Sunspotter, can help determine whether an eruption may be a threat if it is Earth-directed. The Earth’s atmosphere is also monitored to detect changes related to solar wind variations, as well as short-term impacts of solar eruptions. Ongoing scientific research is crucial to determine the fundamental physical processes involved in driving space weather, such as solar magnetic fields. The more that is known about these processes, the more models can be improved to accurately predict when a flare or eruption will occur. So keep clicking those active regions and help improve our warning systems!
*Disclaimer: all statements in this post are my own, and not those of the Met Office.
… Then tell us so!
Why I ❤ Citizen Science
The Sunspotter volunteers have been working tirelessly at classifying all of our sunspot data, and it shows! We are already near 250,000 classifications, making this dataset around 70% completed. After this one is done, we will be releasing a new dataset that includes over 200,000 sunspot group images, and much more meta data.
Also, we have been experimenting with ways to ‘de-project’ the data so that sunspot groups near the edge or ‘limb’ of the Sun do not appear squashed due to foreshortening- but, more on that when we launch the new dataset…
While working on this project, our team has had the privilege of interacting with some of the Sunspotter volunteers. If you haven’t tried it yet, pop over to talk.sunspotter.org and give us a shout! The volunteers taking advantage of Talk have been asking us some fascinating questions about the science behind Sunspotter, why the data looks the way it does, and about our ideas for how the data will be analysed later on. Honestly, it has provoked us to think much more deeply about this project than we would have otherwise.
I think that this is one of the hidden benefits of starting a citizen science project: your grandad’s/grandma’s data analysis algorithm is no longer a lifeless automaton. Now, it is a group of people that can provide you with amazing feedback on the work!
The Cursed Pay-Wall
Over the course of interacting with volunteers on Talk, there have been many discussions about the science behind Sunspotter where a specific research paper came up in the conversation. Unfortunately, many solar and space-physics researchers have not yet discovered the likes of arxiv.org, where you can upload a ‘pre-print’ manuscript of your science journal paper for everyone to read for free. So, often when a Sunspotter volunteer goes to check out a specific paper (perhaps by using the astrophysics-paper search engine, adsabs), they are hit with a ‘Pay-Wall’. If you have not encountered this yet, it occurs when an individual tries to access a science paper on-line, using a computer that is not on a university, research laboratory, or library campus (that has access to the given journal). Often the publisher requires a fee of ~$30 to access a single paper!
It is my personal opinion (and the opinion of many others) that at the very least, the general public (non-working scientists) should be able to freely access any scientific research paper in which the work was paid for by public funding. The vast majority of research is paid for by public money, but the majority of scientific papers published today are not Open Access. On the bright side, there is a growing movement to make science fundamentally freely accessible for everyone.
That said, when a research paper comes up in a Sunspotter conversation, that is not openly accessible, I have taken to emailing the author directly to request that they upload their paper to arXiv to make it available for free. Furthermore, I have begun to email the main astrophysics journal publishers to request that they provide open access to volunteers of citizen science projects. The American Geophysical Union (AGU) has already responded very positively to our request, and is working to provide such access! Furthermore, the folks at Zooniverse got the Monthly Notices of the Royal Astronomical Society (MNRAS) to agree make all citizen science-based research in their journals automatically open-access! Cheers to AGU and MNRAS!
These journals are clearly thinking ahead, and hopefully the other journals will see the light soon!
To help us in our mission of making science more open (especially for citizen science volunteers), please vote in our poll above, to declare your interest (or disinterest) in free access to journal papers. If this is an issue that Zoo-ites feel strongly about, we would like to know. For one thing publishers will take our requests more seriously if we can show that there is substantial support.
Again, we at Sunspotter really appreciate all the time, effort, and interest that you volunteers have given us!
16 Years Staring at the Sun (without sunglasses)
The data we are using for this stage of Sunspotters comes from the Michelson Doppler Imager (MDI; Scherrer, 1995) instrument, which is on-board the Solar and Heliospheric Observatory (fact sheet; shown in images above).
SOHO orbits the Sun between the Sun and Earth at the first Lagrange point (L1). Enjoy a video of SOHO being launched into space on board an ATLAS rocket. An L1 orbit allows uninterrupted views of the Sun, without the Earth or Moon getting in the way. The MDI instrument was turned off in 2011, but successfully took data for about 16 years.
Around 60,000 magnetic images of the Sun’s surface were beamed back to Earth over this time interval, and have allowed the study of the magnetic properties of sunspots and the Sun as a whole over more than an entire 11-year solar cycle. In this project we take advantage of these features to study the magnetic complexity of sunspot groups over a long timescale and with regards to eruptive activity.
The current dataset used in Sunspotter includes cut-out images that are based on the locations of sunspot groups determined by hand (by the National Oceanic and Atmospheric Agency and the US Air Force; Figure 14). The current dataset includes around 10,000 images, and will allow us to determine the relationship between sunspot group magnetic complexity and other magnetic properties, such as magnetic area, flux, polarity imbalance, and the length of the polarity separation line (separating positive and negative regions of a given sunspot group image).
Making Sense of the Data
We have processed the data in a specific way to aid volunteers in comparing the sunspot group images. The thick white line shows the limb of the Sun (beyond the limb lies outer space). MDI provides us with images of the whole disk of the Sun. We have cut out images of sunspot groups centered on a set of human detections, as explained above. The cut-outs end up being all different sizes, so we buffer out the smaller ones with generated noise to make them a uniform size.
Within the images you will see blobs of white and black. White areas represent magnetic fields oriented toward the observer, and black areas represent magnetic fields oriented away from the observer. If you could put a bar magnet on the Sun, the magnet would look white when facing one way, and black when facing the other way.
However, not everything you see in these images is a magnetic field (or necessarily even a physical feature, for that matter). The following images show some examples of odd looking stuff you’ll find in the data. We have tried to explain the cause of each observed feature, below.
You will probably notice that the sunspot groups have a rectangular boundary. This is due to the detection algorithm we used to extract each sunspot group. We get rid of all the stuff that we do not consider to be a part of the sunspot group in question. It is an arbitrary choice of what to keep and what to get rid of, but one that has to be made- at least it is done in a uniform manner. You will be able to see a ‘context’ image of what exists outside of this rectangular area after making a classification and going to the ‘Profile’ tab. We buffer out the images to the same field of view so that one can compare the scale of the features in the images. We are discussing the possibility of removing the scale information in the next batch of data (to remove the potential bias that larger spot groups would always be judged to be more complex- which should not always be the case…).
Very rarely, a scratchy speckled pattern will be seen, like that shown in the image above. This is due to a large eruption being launched from the Sun, that actually hit our spacecraft! The scratches and speckles are energetic particles (e.g., protons), travelling close to the speed of light that hit our camera and were recorded, while we were trying to take a picture of the Sun. Its kind of like trying to film a tornado in Kansas, and your film crew keeps getting hit by airborne cows. Movie of the ‘Halloween Storm’ in 2003; pay attention at ~8 seconds into the movie.
The black line seen at the bottom of the feature in this image is likely due to MDI’s camera not recording certain pixels. Once in a while a large block of data (not shown here) may go missing, and this tends to occur when the ground station on Earth was in the process of receiving data and an interruption occurred, resulting in the loss…like having dropped cell-phone call.
These images show sunspot groups during their emergence phase. The first and last images show pairs of sunspots (one positive/white, and the other negative/black) bursting through the solar surface. They often show a characteristic double-C shape, like two lions roaring at each other, face-to-face. The middle image shows a sunspot pair emerging (in the top left corner of the image) into an already established sunspot group (the stuff in the center of the image). One explanation for the double-C is that it is the result of helical magnetic fields passing through the solar surface (all you can see is the cross-section). Here is the best example I have ever seen, as noted by Sunspotter volunteer, @artman40.
Often, you will notice that circular blobs of a given polarity will have an edge that appears to be of the opposite polarity. It will always be the edge facing the limb (edge) of the Sun. These circular blobs are sunspots, and because their magnetic fields fan out at their boundaries (penumbrae), this can cause a ‘false’ polarity to be observed when the sunspot is near the limb of the Sun, because of the way that we measure magnetic fields with this instrument. The ‘true’ polarity of a sunspot is the one that is on the edge facing away from the limb of the Sun. In the images above, the first sunspot is ‘truly’ positive/white, the middle one is ‘truly’ positive/white, and the last one is ‘truly’ negative/black.
The ‘hollowed-out’ artifact seen in the leading sunspot is due to a saturation effect, as noted by Sunspotter volunteers @Quia and @Mjtbarrett. The problem is that for MDI, much of the data was processed on-board the spacecraft. Because of the way that the processing was designed from the beginning, the model used to calculate the magnetic fields broke down for very large fields (>2000 gauss), resulting in the saturation. Unfortunately, the effect isn’t even linear, so even at lower fields (1k-1.5k G) you start to see problems. Also, the non-linearity makes it hard to correct for.
A paper on the effect is available. In a future blog post, we will explain the making of a magnetograms, and also touch on the saturation problem.
Often, there will be nothing to see at all. This is usually because the sunspot group that was recorded by human observers at the beginning of the day has already progressed beyond the edge of the Sun by the time our spacecraft took the data. And since we are relying on human observers to pick out sunspot groups in the current data set, sometimes we end up with nothing…
Last but not least: this is an image of a particularly large sunspot group that released many significant eruptions (some of the largest that we know of!). Note the elongated strip of negative (black) magnetic flux, sandwiched between the two areas of positive (white flux). This sandwiching may have caused the shearing and stretching of the fields to the point of breakage. The double polarity separation line pattern seen here (tracing the dividing line between the white and black areas) is of particular interest to me, as I am convinced it that it is often a sign that a large eruption could occur.
If you come across any other interesting artifacts or patterns in the data, save them by going to the ‘Profile’ tab in the main sunspotter interface. There is plenty to discover about what makes sunspots go boom!