A solar flare occurs when the solar atmosphere suddenly releases the magnetic energy that has been built inside. We can say that a solar flare happens by the eruption of the magnetic fields of sunspots. A solar flare contains high energy particles such as electrons, protons, and heavy nuclei; and the release time of the flare is relatively short- just a few minutes¹.

Sunspots are the appearances of the places on the Sun’s surface where the powerful magnetic fields break and extend into the Sun’s atmosphere. Between sunspots with opposite magnetic poles, coronal loops often arch². Coronal loops are bright curves on the Sun that come in many sizes. They are the visible formations of the magnetic field on the surface of the Sun. When these loops touch, they create a short circuit, which triggers the explosions. The amount of energy released is the equivalent of millions of 100-megaton hydrogen exploding at the same time¹. This process is called “magnetic reconnection”. A magnetic reconnection is the breaking and reconnecting of the magnetic field lines which are opposite to each other in a plasma. During the reconnection, magnetic field energy is converted to kinetic and thermal energy. This conversion causes the particles to be charged with high energy and ejected into space³.

Picture: A magnetic loop, or prominence on the Sun¹.


The actions happening during the flare are grouped in 3 steps. These are the stages of a solar flare.

The first step is called the “precursor stage”, in this phase the release of magnetic energy is triggered. Soft X-ray emission is detected here. Soft X-rays have lower energy than normal X-rays and are more easily absorbed by air and other environments such as water or some solid materials⁸.

The second step is the “impulsive stage”, where protons and electrons are accelerated by high energies. Radio waves, hard X-rays, and gamma rays are emitted in this phase.

The third step is called the “decay stage”. The gradual build-up and decay of soft x-rays can be detected in this phase⁷. We can say that flare radiation can be observed in every wavelength, from radio waves to gamma rays during the solar flare stages.

It is found by NASA’s Solar Dynamics Observatory (SDO) that about 1 in 7 flares experience an “aftershock”. An aftershock is the revival of the solar flare 90 minutes after its death. When the flare revives, it produces an extra fluctuation of extreme ultraviolet radiation. This late phase happens when some of the sunspots’ magnetic loops re-form⁴.

Picture: The Sun in soft X-rays. On the right part of the Sun, the bright part, there are post-flares. Post flares are the hot ones that remain after the flare⁵. The sources of the x-rays are shown with a graphic.


In the classification system that divides the solar flares according to their strength, the smallest ones are “A-class flares”, followed by B, C, M, and X. Each letter represents a 10 fold increase in energy output. So X is ten times an M and 100 times a C. And each class has 9 more different strength levels inside them- except X.

C-class and smaller flares are not strong enough to noticeably affect Earth. M-class flares can cause minor radiation storms that might damage astronauts.

Another type of solar flares is X-class flares. It is divided into many levels inside. There are flares more than ten times more powerful than X1, therefore X-class flares can go much higher than 9.On the 28th of October in 2003, the most powerful flare was measured with modern methods and it was so powerful that it even managed to overload the sensors measuring its power. The sensors cut out at X28⁶. Later analysis found that the flare reached a peak strength of about X45, NASA has said¹¹. It was, in fact, one hundred quadrillion times more powerful than the level that sensors had found.

Picture: The photo of the 2003 flare. The solar storm was part of a sequence of at least nine major flares over a two-week period¹¹.


In 1859, a 33-year-old astronomer Richard Carrington traced the outlines of a large group of sunspots while projecting an image of the Sun onto a screen and sketching what he saw. Then suddenly he saw two brilliant beads of white light appear over the sunspots. And that was the first sun flare that any human had ever seen.

Picture: Sunspots sketched by R. Carrington on Sept. 1, 1859. © R. Astronomical Society⁴.

Since then many solar flares were observed through various instruments from basic telescopes to the most complex spectrometers. One of the other solar events¹¹ is the 1972 flare, which was so powerful that AT&T, which is an international telecommunication company, needed to redesign its power system for transatlantic cables after the event, as NASA said.

Another flare worth mentioning is the 1989 flare. It set off a major energy blackout in Quebec, Canada and six million people was left out without electricity for 9 hours on the 13th of March.

Another event is 2000 The Bastille Day Event. It takes its name from the French national holiday. It was a large flare classified as an X5 level. It is still one of the most highly observed solar storm events and was the most powerful flare since 1989.

A bigger flare happened in 2006 Dec. 5. It was an X9 level flare and it managed to damage the solar X-ray imager instrument on the GOES 13 satellite that took its picture, NOAA officials said. Solar flares are one of the most studied phenomena in astronomy⁴.


In order to understand the differences between these two astronomic events, we need to know what a CME is. CME’s are strong eruptions that happen near the Sun’s surface and surge through the Solar System. During a CME, giant bubbles of plasma come out from the Sun. The plasma is a superheated cloud of protons and electrons carried by the solar wind. In several hours billions of tons of material leave the Sun’s surface and accelerate to speeds of a million miles per hour⁹.

Picture: A CME was taken in 2000. The Sun is blocked by a disc in order to prevent the light. The white circle indicates the Sun’s surface⁹.

Although both eruptions are created when the motion of the sun’s interior makes its own magnetic fields twist, the appearance of the explosions is different. The flare can be seen anywhere in the vicinity like a flash; however the CME is like a cannonball, it is shoved to a single target and only affects there. Solar flares appear as a bright light and CMEs appear as enormous fans of gas swelling into space.

Another difference is their effects on Earth. The energy of a flare can distort the area of the atmosphere where radio waves travel. It can damage communication like it was mentioned before.

On the other hand, CMEs can funnel particles into near-Earth space. A CME can create flows that drive particles down toward Earth’s poles by disturbing the magnetic fields of the Earth. When these particles react with oxygen and nitrogen, they help create the aurora, also known as the Northern and Southern Lights¹⁰.

Prediction of space weather, or specifically solar flares, in this case, has great importance due to its effects on space programs, satellites, even air traffic and, a big flare is highly dangerous for our society¹². Solar flares do not follow a specific pattern which makes it harder to predict. However, some models such as¹⁶ Wang-Sheeley-Arge (WSA) solar model, a geometric coronal mass ejection (CME) cone model, and the Enlil ambient solar wind model can be used in order to make better predictions. Since it is almost impossible to see the end of the tunnel, the only thing that can be done is to make some guesses about what the data means. Over the years, scientists came up with solutions to this crucial problem. Prediction of solar flares is important due to its broad range of aforementioned effects. One of the ways to predict solar flares is to use machine learning (ML). Machine learning is a field associated with artificial intelligence (AI) which allows computers to learn from a set of data and apply them and predict future data¹⁴. The reason why ML is beneficial is that it helps us to process a really big amount of data that humans cannot process. ML can be used to create algorithms that can learn the data and make decisions from them. One of the important parameters to use to predict solar flares is the X-ray flux which shows us the state of the ionosphere, observed by the Geostationary Operational Environmental Satellite (GOES) series. Also, it is widely used to assess radio wave propagation¹³,¹⁵. With UFCORIN¹⁷, to predict space weather, they followed these steps:

  • Read a prediction strategy that has a list of time series data. Load the input and output time series from solar observation archive data
  • Separate the data in order to perform cross-validation
  • Conduct optimized regression predictor of the model function from the training set data
  • Prediction of the output for the test data
  • Predict the flare events
  • Compare the prediction and observation and generate contingency tables
  • Calculate true skill statistics

Another important parameter to consider is the cone model. The cone model is a geometrical analysis for CMEs that are three-dimensional and represented as a cone. Through this model, in forecasting space weather, the primary means are to determine kinematic and morphological data of CMEs¹⁶,¹⁹.

Many researchers demonstrated that using photospheric vector magnetograms in combination with ML can predict solar flares effectively¹⁴. There are several algorithms that are used in ML, i.e., decision trees, neural networks. Also, it is important not to skip the importance of the cross-validation method which trains the algorithm on a subset and tests it on another subset¹⁵.

Locally Weighted Regression (LWR) is a “non-parametric, memory-based technique that stores past data exemplars in memory and process them when a new query is made”. When a query is made, the algorithm locates similar examples and practices the LWR process, a weighted regression with the local observations. We can say that there are differences between the techniques used in Locally Weighted Regression and the parametric techniques, e.g. neural networks. Also, another definition of LWR can be that it is a non-linear, non-parametric extension of linear regression²¹. LWR is an alternative model to estimate true model variance, a regression model is obtained for each value of the predictor where varied weights are assigned to all observations which help to find local patterns that parametric regression misses¹⁸. Techniques in the regression model have a regression fitted to the data, then, with testing the anomaly score is determined. However regression model is useful when training data is available²⁰. With showing that the LWR model is fast and “fairly accurate”, it is also shown that the LWR model has some advantages over neural network methods where training time is longer and model updates are harder²¹.

On our planet, Earth, there are many issues that affect the climate and meteorology. One of these issues is the Sun itself. Since space weather mostly depends on the things happening in the Sun, solar flares are one of the events occurring there and affecting our planet. The most important the Sun has on Earth is the irradiance or the brightness of the Sun²². Irradiance can be defined as “the energy emitted or the number of photons emitted per unit area per time²⁵”. The units used for irradiance are W 1/(m²) x [1/ (wavelength bin)] or photon 1/(s x cm²)(wavelength bin)-1 where the wavelength bin can be defined any size²⁵. Depending on the wavelength of the photons coming from the Sun, it has a great impact on absorbing the energy by ozone and the stratosphere. The minimal effects are caused by the Extreme Ultraviolet wavelengths. In his paper, Pawlowski showed that solar flares can affect the thermospheric density in which it enhances²³. Even though the Sun affects the thermosphere, it does not necessarily mean that it causes immediate effects on meteorology and climate changes. However, in the long run, solar flares might cause climate change²⁴.


Being a magnificent natural phenomenon, solar flares may also have various negative effects on technology, satellites, and space tasks; especially when they happen in X-class. Although coronal mass ejections are more likely to cause more damage since they carry more material to a larger area of space, increasing the possibility of the material interacting with the Earth²⁹, effects of solar flares are worth mentioning as there are a few big solar flare events in history where the harmful effects are clearly represented.


Abnormally strong solar flares can actually cause temporary blackouts and create electrical currents on the ground that damage the electrical utilities by disturbing the Earth’s magnetic field. When an electrical current is created by the disruption of the Earth’s magnetic field by space weather, it is called Geomagnetically Induced Currents (GIC)³³. The generation of GIC is explained by Faraday’s law of induction. Faraday’s law says whenever there is a change in a magnetic field, it creates currents in conductors and an electric field emerges.

The previously mentioned Quebec Event is a perfect example for understanding the potential damage on technology. On the evening of March 12, 1989, a giant cloud of solar plasma entered the Earth’s magnetic field, which created an intense magnetic disturbance. This caused electrical currents to happen in the ground under much of North America and on March 13, the entire Quebec grid lost its power. Daily life in Canada was paralyzed, children couldn’t go to school, transportation was shut down and people could not do their jobs, millions of people were left out in the darkness.

Some of the U.S. electrical utilities had problems too. It was reported that across the United States about more than 200 power grid problems were encountered. New York Power lost 150 megawatts and The New England Power Pool lost 1,410 megawatts of energy. Service to 96 utilities in New England was interrupted. The U.S. barely had the spare energy to use during the time²⁶.

There were three main reasons why the Quebec grid was heavily affected³⁴: the length of the transmission network, lack of a device to reduce GIC flows and the incorrect settings of Static VAR compensators (SVCs), which is an electrical device used for creating or absorbing reactive power for the electrical power systems. The large GIC flow in the system caused SVCs to trip. Later, it was found that the SVCs were working way below their capacities. Their overvoltage and overcurrent limits were increased and they were able to easily withstand the following magnetic events.

Another similar event is the Carrington Event, the first known solar event in history that happened on August 28, 1859. That day, telegraph communications around the world began to fail and many people reported the sparks coming from the telegraph machines and even some telegraph stations that used chemicals to mark sheets reported the sheets burning because of the powerful surges. Many telegraph lines across North America were severely damaged by the first two sequential solar storms. In a telegraph service in Pittsburgh, the currents flowing through the wires resulting from the solar storms were reported to be so powerful that it left the platinum contacts under the danger of melting and streams of fire coming out the circuits could be seen³¹.


Solar flares can cause a satellite to become corrupted, lose the signal connection, or lose its orbit.

Magnetic storms on the Sun can harm the Earth-orbiting satellites, especially the higher ones which are usually used for communications. The satellite can become highly charged and a component of it becomes damaged by the high current that was poured into the satellite. Or the component can be damaged by the high energy particles passing through the satellite²⁹.

Solar flares can also cause radio signals to fade. Research made in 2001, shows the effect of the solar flares on signals³⁰. The majority of the fadings are caused by the x-class flares as expected.

Picture: Examples of fading events associated with C-class (panel i), M-class (panel ii) and X-class (panels iii and iv) solar flare events. Each panel shows: (a) average signal strength and the received signal strength, (b) X-ray solar flux and © the amount of absorption, which is defined as the difference between the observed signal strength and average signal strength. It is easily noticed that whenever x-ray flux increases, the amount of received signal decreases like it is responding.

Earth’s atmosphere can be heated up by the high-energy particles and radiation from the Sun. These particles collide with the molecules like nitrogen and oxygen in the atmosphere and heat energy is created. The heated air rises and the atmosphere expands. If the expansion is too much, it will eventually reach the lower-orbiting satellites and create a resistance to them. This results in a drop in the altitude of the satellite, which is called orbital drag²⁷. As mentioned in the November issue of Space Weather, during an extreme magnetic storm event, a satellite could drop nearly a third of its height in one day, which shows the importance of this drag effect. Also, NASA has stated that the amount of the damage caused by the solar flare is more closely correlated by the duration of the flare than the strength of it.

Satellites don’t fall out of the sky because of the orbital drag but their communication with the Earth is disrupted, their lifespans are shortened and the possibility of collisions and getting lost in space is increased by the orbital drag. So NASA scientists are careful about the space weather and orbital drag since low-Earth satellites are primarily for the telecommunication systems and for weather observations. Here is a picture of the magnitude of the lost satellites due to the orbital drag.

Picture: Number of satellites lost in connection with the March 13–14, 1989 storm, the Quebec Event (UCAR)²⁸.

International Space Station is also being affected by orbital drag. ISS travels on a 51.6 degrees tilted orbit. As the ISS travels from west to east, the altitude of the orbit slightly drops over time due to the gravitational pull and the atmospheric drag combined. So the astronauts of ISS have to regularly observe the solar storms and periodically adjust its orbit with the reboosts³².

Picture: The change of the orbit of ISS is shown here³².

Fortunately, the astronauts working there don’t have to worry about the radiation caused by the flares all the time as they stay relatively close to the Earth- about 400 kilometers. However, during space walks or missions to the Moon or Mars, cumulative radiation exposure can be dangerous to an astronaut²⁹.


Nurperi Berfin İskenderoğlu and Beril Bilici.


  1. Space Environment. What is a solar flare? [accessed 2021 Jan]. https://www.qrg.northwestern.edu/projects/vss/docs/space-environment/3-what-is-solar-flare.html#:~:text=A%20solar%20flare%20occurs%20when,each%20other%2C%20setting%20off%20explosions.&text=When%20these%20loops%20hit%20each%20other%2C%20they%20make%20a%20solar%20flare
  2. Coronal Loops in the Sun’s Atmosphere. Coronal Loops in the Sun’s Atmosphere | UCAR Center for Science Education. 2014 [accessed 2021 Jan]. https://scied.ucar.edu/sun-coronal-loops#:~:text=Coronal%20loops%20are%20often%20%22rooted,corona%2C%20the%20Sun's%20upper%20atmosphere
  3. What is Magnetic Reconnection? [accessed 2021 Jan]. https://mrx.pppl.gov/Physics/physics.html#:~:text=Magnetic%20reconnection%20(henceforth%20called%20%22reconnection,events%20in%20our%20solar%20system
  4. Phillips T. The Secret Lives of Solar Flares. NASA. 2011 Sep 19 [accessed 2021 Jan]. https://science.nasa.gov/science-news/science-at-nasa/2011/19sep_secretlives
  5. Coronal Hard X-ray Sources in Solar Flares. [accessed 2021 Jan]. http://www.issibern.ch/teams/CornSource/
  6. Fox KC. Solar Flares: What Does It Take to Be X-Class? NASA. 2013 Jul 31 [accessed 2021 Jan]. https://www.nasa.gov/mission_pages/sunearth/news/X-class-flares.html#:~:text=Solar%20flares%20are%20classified%20according,M%20and%20X%2C%20the%20largest.&text=Solar%20flares%20are%20giant%20explosions,high%20speed%20particles%20into%20space
  7. Overview of Solar Flares. NASA Goddard Space Flight Center. [accessed 2020]. https://hesperia.gsfc.nasa.gov/rhessi3/mission/science/overview-of-solar-flares/index.html#:~:text=There%20are%20typically%20three%20stages,million%20electron%20volts%20(MeV)
  8. 7 things you may not know about X-rays. Argonne National Laboratory. 2013 Sep 13 [accessed 2020]. https://www.anl.gov/article/7-things-you-may-not-know-about-xrays
  9. Crockett C. What are coronal mass ejections? EarthSky. 2020 Dec 9 [accessed 2020]. https://earthsky.org/space/what-are-coronal-mass-ejections
  10. Gleber M. The Difference Between Flares and CMEs. NASA. 2014 Sep 21 [accessed 2020]. https://www.nasa.gov/content/goddard/the-difference-between-flares-and-cmes
  11. Malik T. The Sun’s Wrath: Worst Solar Storms in History. Space.com. 2013 Aug 7 [accessed 2020]. https://www.space.com/12584-worst-solar-storms-sun-flares-history.html
  12. O’Callaghan J. We can now predict dangerous solar flares a day before they happen. 2020 Jul 30 [accessed 2020]. https://www.newscientist.com/article/2250369-we-can-now-predict-dangerous-solar-flares-a-day-before-they-happen
  13. Tlatov AG, Illarionov EA, Berezin IA, Shramko AD. Prediction of Solar Flares and Background Fluxes of X-Ray Radiation According to Synoptic Ground-Based Observations Using Machine-Learning Models. Cosmic Research. 2020;58(6):444–449. doi:10.1134/s0010952520060106
  14. 1. Liu H. Machine Learning for Scientific Data Mining and Solar Eruption Prediction. Ann Arbor: New Jersey Institute of Technology; 2020.
  15. Ahmed OW, Qahwaji R, Colak T, Higgins PA, Gallagher PT, Bloomfield DS. Solar Flare Prediction Using Advanced Feature Extraction, Machine Learning, and Feature Selection. Solar Physics. 2011;283(1):157–175. doi:10.1007/s11207–011–9896–1
  16. 1. Murphy JJ. Advanced analysis and visualization of space weather phenomena. Ann Arbor: University of Colorado at Boulder; 2017. 147 p; Last updated — 2020–11–26.
  17. Muranushi T, Shibayama T, Muranushi YH, Isobe H, Nemoto S, Komazaki K, Shibata K. UFCORIN: A fully automated predictor of solar flares in GOES X‐ray flux. Space Weather. 2015;13(11):778–796. doi:10.1002/2015sw001257
  18. 1. Szychowski JM. Locally weighted regression for improved variable selection. Ann Arbor: The University of Alabama; 2006. 106 p; Last updated — 2020–11–02.
  19. Xie H. Cone model for halo CMEs: Application to space weather forecasting. Journal of Geophysical Research. 2004;109(A3). doi:10.1029/2003ja010226
  20. Carlton AK. Fault Detection Algorithms for Spacecraft Monitoring and Environmental Sensing [thesis]. 2016.
  21. Hines JW, Townsend LW, Nichols TF. SPE dose prediction using locally weighted regression. Radiation Protection Dosimetry. 2005;116(1–4):131–134. doi:10.1093/rpd/nci010
  22. Space Weather Impacts On Climate. Space Weather Impacts On Climate | NOAA / NWS Space Weather Prediction Center.
  23. Pawlowski DJ. On the response of the upper atmosphere to solar flares. Ann Arbor: University of Michigan; 2009. 152
  24. Do solar storms cause heat waves on Earth?: NOAA Climate.gov. Do solar storms cause heat waves on Earth? | NOAA Climate.gov. 2012 Apr 12. https://www.climate.gov/news-features/climate-qa/do-solar-storms-cause-heat-waves-earth
  25. 1. Rodgers EM. Solar flare soft X -ray irradiance and its impact on the earth’s upper atmosphere. Ann Arbor: University of Alaska Fairbanks; 2007. 204 p
  26. Odenwald S. The Day the Sun Brought Darkness. NASA. 2009 Mar 13 [accessed 2021]. https://www.nasa.gov/topics/earth/features/sun_darkness.html
  27. Johnson-Groh MJ-G. Past Solar Superstorms Help NASA Scientists Understand Satellite Risks Tran L, editor. NASA. 2020 Nov 30 [accessed 2021 Jan]. https://www.nasa.gov/feature/goddard/2020/solar-superstorms-past-help-nasa-scientists-understand-risks-for-satellites-orbital-drag
  28. Satellite Drag | NOAA / NWS Space Weather Prediction Center. Noaa.gov. [accessed 2021]. https://www.swpc.noaa.gov/impacts/satellite-drag
  29. The Impact of Flares. National Aeronautics and Space Administration Goddard Space Flight Center. 2019 [accessed 2021]. https://hesperia.gsfc.nasa.gov/rhessi3/mission/science/the-impact-of-flares/index.html
  30. Contreira DB, Rodrigues FS, Makita K, Brum CGM, Gonzalez W, Trivedi NB, da Silva MR, Schuch NJ. An experiment to study solar flare effects on radio-communication signals. Advances in Space Research. 2005 [accessed 2021 Feb 7];36(12):2455–2459. doi:10.1016/j.asr.2004.03.019
  31. Klein C. A Perfect Solar Superstorm: The 1859 Carrington Event. HISTORY. 2012 Mar 14 [accessed 2021]. https://www.history.com/news/a-perfect-solar-superstorm-the-1859-carrington-event
  32. Evans CA, Robinson JA. Space Station Orbit Tutorial. Gateway to Astronaut Photography of Earth. [accessed 2021]. https://eol.jsc.nasa.gov/Tools/orbitTutorial.htm
  33. Pirjola R. Geomagnetically induced currents during magnetic storms. IEEE Transactions on Plasma Science. 2000 [accessed 2021];28(6):1867–1873. doi:10.1109/27.902215
  34. 9. Guillon S, Toner P, Gibson L, Boteler D. A Colorful Blackout: The Havoc Caused by Auroral Electrojet Generated Magnetic Field Variations in 1989. IEEE Power and Energy Magazine. 2016 [accessed 2021];14(6):59–71. doi:10.1109/mpe.2016.2591760

Koç Üniversitesi Bilim Kulübü Resmi Medium Hesabı | Official Medium Account of Koç University Science Club

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store