Coronal Mass Ejections: a Driver of Severe Space Weather

Coronal mass ejections: a driver of severe space weather

Lucie Green and CME being formed per day at cycle mini- the CME with the magnetosphere. If this 1 No. 70, Vol. 2015, January – Weather mum and as many as five per day at cycle occurs, the dayside magnetospheric flux Deb Baker maximum (Webb and Howard, 2012). The gets dragged over to the nightside by the Mullard Space Science Laboratory, UCL solar cycle is driven by an evolving global motion of the CME, and an accumulation magnetic field that changes in strength and of magnetic flux then occurs in this part of complexity and which manifests itself in the the magnetosphere, followed by magnetic Introduction photosphere by the presence of sunspots – reconnection. This reconfiguration of the Observational studies of the solar atmos- regions of concentrated magnetic field. As nightside magnetosphere allows magnetic phere in the early 1970s revealed that the the solar cycle ebbs and flows, the magnetic flux to move back to the dayside, where Sun sporadically ejects vast quantities of field in the solar atmosphere is transformed it can undergo reconnection with the CME matter into the Solar System in addition between a magnetically simple state when once again. This process is known as the to the constant outflow known as the there are correspondingly few sunspots at Dungey cycle (Dungey, 1961). During these solar wind (Tousey, 1973). These sporadic the surface, and a magnetically complex times the magnetosphere is driven into a ejections are known today as coronal configuration when there are many sun- disturbed state, producing geomagnetic mass ejections (CMEs). They are seen in spots. In this way, CMEs can be understood sub-storms and storms which then lead images that use an occulting disc to block as a mechanism by which the Sun ejects to a multitude of space weather effects the dazzling light emitted by the photo- atmospheric magnetic field that harbours (Pulkkinen, 2007). sphere, creating an artificial solar eclipse both magnetic energy and a complex mor- CMEs, which are the bulk ejection of and allowing the outer atmosphere to be phology (Low, 1996). CMEs play an impor- magnetised plasma, are often temporally viewed (Figure 1). CMEs are then revealed tant role in the magnetic cycle of the Sun. and spatially associated with another form as outward moving structures that exhibit For recent review articles on CMEs see, for of solar activity known as a solar flare – a range of morphologies and which carry example, Webb and Howard (2012), Chen sudden bursts of electromagnetic radiation around 1012kg of magnetised plasma into (2011) and Forbes et al. (2006). (decametre radio waves to gamma-rays) the heliosphere with an average speed and high-energy particles (Benz, 2008). of ~490kms−1 (Webb and Howard, 2012). This CME-flare association has its origins Upon eruption, these magnetic structures CMEs and their relevance to in the physical processes that are common quickly expand to become larger than the space weather to both phenomena. Solar flares arise from Sun itself. The discovery of CMEs goes beyond elu- the conversion of magnetic energy into CMEs are common events and occur with cidating important aspects of the Sun’s other forms during magnetic reconnec- a frequency that follows the (on average) magnetic field evolution. The discovery of tion, and this energy conversion leads to 11-year solar cycle with approximately one these ejections answered an outstanding the production of electromagnetic radia- question in solar-terrestrial physics when it tion. However, magnetic reconnection also was realised that they are the solar phe- reconfigures the magnetic field during a nomenon that are the cause of major, but CME, which accelerates the magnetised transient, disturbances to the geomagnetic plasma away from the Sun. Despite these field (e.g. Gosling, 1993). When CMEs collide common aspects, CMEs and flares are very with the Earth’s magnetic field (the magnet- distinct in terms of their physical guise. Solar osphere) they can compress its dayside and, flares have a space weather impact through if the magnetic field of the CME is of the their high-energy radiation that ionises the correct orientation, magnetic reconnection Earth’s upper atmosphere, and through the between the CME and the magnetosphere protons that are accelerated to near-relativ- can occur. Magnetic reconnection is a fun- istic energies, posing a threat to spacecraft damental physical process in which plasma and humans in space. and magnetic field become momentarily Figure 1. A SOHO/LASCO image of a coronal decoupled, leading to a reconfiguration of mass ejection that occurred on 2 June 1998. Extreme space weather events The erupting structure is seen in the lower the field and the conversion of magnetic right and exhibits a twisted or helical shape. energy into other forms of energy. The mag- The most extreme space weather event in The white circle indicates the size of the solar netic field at the sub-solar point of the mag- our records is the geomagnetic storm that disc; the size of the occulting disc is shown by netosphere (directly between the Sun and began on 2 September 1859 (Tsurutani the blue filled circle. The SOHO spacecraft is the Earth) is strongly northward, so a strong et al., 2003). This storm has become known located at the L1 position. (Image courtesy of southward directed magnetic field within as the Carrington event, named after the ESA/NASA - http://sohowww.nascom.nasa.gov/ the CME could lead to magnetic reconnec- Victorian amateur astronomer who observed gallery/images/c2helix.html) tion and a joining of the magnetic field of the associated flare, which occurred in a 31 large sunspot group he was drawing at his One thing remains clear though: the drag force in the solar wind. In addition, it observatory in Redhill (Surrey). Less than a most intense geomagnetic storms will not needs to hit the magnetosphere head-on day later, an intense display of the aurora be triggered unless the CME has the cor- (i.e. not give us a glancing blow), have a erupted, which was seen down to excep- rect magnetic field configuration. Intense high dynamic pressure and have a magnetic tionally low latitudes as an intense geomag- geomagnetic storms require the presence configuration with a strong and sustained netic storm took hold. Measurements from of a strong southward magnetic field southward component. It is this combina- a magnetometer in Bombay, India, revealed component more than any other plasma tion of factors that needs to be accurately the exceptional changes to the Earth’s mag- parameter, (Gonzalez et al., 2011) and this identified in a CME from the plethora that netic field that occurred (Siscoe et al., 2006). southward field must persist for several leave the Sun. The significant aspect of this event, which hours (Russell et al., 1974). This is because The propagation time of CMEs means that was unknown to Carrington at the time, the strong southward component provides forecasting their occurrence may not neces- was that the sunspot group had produced an effective coupling of the magnetic field sarily be required. In the majority of cases, Coronal mass ejections a CME at the same time as the flare. The CME of the CME with the magnetosphere, which space weather users have upwards of a day reached the Earth after only 17.5h, giving it controls the transfer of energy. NASA’s ACE to make their preparations once an eruption a mean velocity along the Sun–Earth line of spacecraft at the first Lagrange point con- has been observed. Once a CME has been ≈2300kms−1. The most intense geomagnetic stantly monitors this magnetic field com- observed, models exist to forecast the direc- storms (Gonzalez et al., 1994) are associated ponent, 1.6 million km upstream of the tion of propagation and, if Earth-directed,

with CMEs that have a large velocity as this Earth. It is referred to as the BZ component the CME arrival time at 1AU (Cargill and can lead to a large inflow of magnetic flux. as its axis is parallel to the ecliptic pole Schmidt, 2002; Owens and Cargill, 2004). Multiple CMEs may also be important as ear- in the Geocentric Solar Ecliptic coordinate However, for a Carrington-type event the lier eruptions could create the solar wind system. CME transit timescale may well be insuf- conditions necessary for a subsequently fast ficient for producing a timely forecast once CME. It looks likely that the Carrington CME Predicting extreme space the delay in receiving the observational occurred at a time of heightened solar activ- data has been factored in. In light of this, Weather Weather – January 2015, Vol. 70, No. 1 ity, and the geomagnetic records suggest weather attempts are being made to determine the another CME had occurred just a few days The role of CMEs in driving extreme geo- likelihood of a CME occurring in the solar before (Boteler, 2006). magnetic activity has led to efforts in both atmosphere based on observations. A brief Intense geomagnetic storms are also modelling and observation to predict their summary of eruptive indicators and time related to CMEs that have a high dynamic velocity and arrival times at Earth. During scales is given in Table 1. pressure as this leads to a high momen- their eruption, CMEs are rapidly acceler- Currently, the longest observational warn- tum flux (Svalgaard, 1977). The great geo- ated at low altitudes in the solar atmos- ing of an Earth-directed CME occurrence is magnetic storm of 13 March 1989 (Allen phere to velocities across a broad range provided by the presence of a so-called sig- et al., 1989) is an example of an extreme from 100kms−1 to more than 3000kms−1. The moid – an S-shaped structure seen in soft geomagnetic storm that was driven by a travel time of CMEs from the Sun to 1AU X-ray or EUV images of the solar atmosphere CME that was not exceptional in terms (1AU is the astronomical unit for the Sun- (Canfield et al., 1999). Once this shape is of its velocity; its mean Sun-Earth tran- Earth distance) varies from less than 24h to seen, the region is likely to erupt within sit time was 770kms−1 (Feynman and more than 3 days; a surprisingly small range, hours (i.e. not days) (Green and Kliem, 2014). Hundhausen, 1994). However, Feynman given the distribution of initial velocities. Due to projection effects, these features are and Hundhausen (1994) remark that the This is explained by the fastest CMEs being best seen when the magnetic configuration CME was exceptionally bright in the corona- decelerated in the solar wind (Woo et al., is viewed from above, that is, when they graph data, which strongly supports an 1985; Watari and Detman, 1998) and the are near Sun centre. Eruptions from these interpretation that it was a very massive slower ones being accelerated by the solar features are therefore likely to be Earth- eruption (the coronagraph images detect wind flow (Lindsay et al., 1999) so that by directed. This is in contrast to another photospheric light scattered by coronal the time CMEs arrive at 1AU the majority observational indicator of an ensuing CME, electrons, and brightness is therefore a are travelling at a speed close to that of namely streamer blowouts, which are seen function of density) and therefore could the ambient solar wind flow (between 350 on the solar limb and are not so likely to be have had a high dynamic pressure when and 550kms−1, Gopalswamy et al., 2000). Earth directed. it reached the magnetosphere. Indeed, the So, for an extreme event the CME needs to Identifying and modelling the pre-erup- dayside magnetosphere is thought to have have a high velocity at 1AU, which means it tion magnetic configuration of sigmoids been compressed towards the Earth, by needs to have a high velocity when leaving using solar observations raises the opportu- more than a factor of two, for an extended the Sun and experience a low aerodynamic nity of inserting this magnetic configuration period of time (Allen et al., 1989). In addition to the above, it is possible that the pre- Table 1 existing condition of the magnetosphere plays a role in severe geomagnetic storms. Summary of the three main observational indicators that a CME is likely to happen along There are studies that indicate precondi- with their timescales. tioning may not play a significant role in Solar feature Eruption indicator References strengthening the storm intensity, but it Filament: Filaments darken and broaden tens Martin (1980) may lengthen the storm duration (e.g. Xie observed anywhere of minutes prior to their eruption. et al., 2008). on the Sun The factors that are important for a Streamer: Streamers brighten over one to Sheeley et al. (1982) and geo-effective CME are the strength of the observed at the several days before eruption. Illing and Hundhausen southward magnetic field component (BZ), limb only They often also include a filament (1986)

the velocity of the solar wind (VSW) and the eruption. solar wind dynamic pressure (ρV ): SW Sigmoid: Their appearance precedes an erup- Canfield et al. (1999) and most easily observed tion by a few hours. Green and Kliem (2014) B V (ρV ) 32 Z SW SW near Sun centre into the propagation models mentioned observations of the solar atmosphere in soft in the solar atmosphere is used to identify above and eventually achieving data-driven X-ray and EUV emission. This identification a realistic configuration. Many studies have simulations of CMEs from Sun to Earth in the allows us to make a direct link between the been carried out using this technique, and coming years. It is also worth noting that, CME at the Sun and the flux rope measured they have found weakly twisted flux ropes since there is a close association between in situ at 1AU. The eruption of a flux rope in in the pre-eruptive magnetic field (Bobra CMEs and solar flares, the use of sigmoids the solar atmosphere is another key focus of et al., 2008; Su et al., 2009; Savcheva and van to predict the occurrence of an eruption is CME research. Their loss of equilibrium due Ballegooijen, 2009). Second is the approach ejections mass Coronal also applicable to efforts in the area of flare to an ideal magnetohydrodynamic instabil- that generates a time-series of 3D coronal forecasting. ity, catastrophe or force imbalance has so magnetic field configurations using obser- There is currently a major gap in our fore- far been well studied and modelled in the vations of the Sun’s surface magnetic field casting capability though. We still do not idealised case (e.g. Hood and Priest, 1981; (Mackay et al., 2011); as the magnetic field have a model that uses solar observations Forbes and Isenberg, 1991; Kliem and Török, evolves, flux ropes can form (Gibb et al., to build a realistic magnetic configuration 2006; Mackay and van Ballegooijen, 2006; 2014). Third, magnetic extrapolations are of a CME as it leaves the Sun. Such a model Démoulin and Aulanier, 2010). Simulations also used to reconstruct the magnetic field would enable us to understand CME evo- of the formation of an idealised flux rope from observations of the three components 1 No. 70, Vol. 2015, January – Weather lution through the inner heliosphere and and its subsequent eruption have also been of the vector magnetic field measured in the therefore forecast the strength of the BZ field achieved (Amari et al., 2010; Aulanier et al., photosphere (e.g. Valori et al., 2012), but as component at 1AU. This is the current space 2010). yet the identification of flux ropes using this weather challenge. However, the first steps Observations of flux ropes prior to their technique remains a challenge. towards achieving this have been taken eruption are crucial for future space weather in fundamental solar and solar-terrestrial forecasting as they can be used as the research. input to create more realistic data-driven Case study: solar active region CME simulations. Flux ropes are modified NOAA 9077 Understanding the magnetic during their eruption as this involves mag- As an example of the promise of the above netic reconnection within the overlying/sur- flux rope models, we briefly discuss here configuration of CMEs rounding magnetic field. This reconnection a CME productive region that was on the When CMEs are observed in situ in the inner adds more magnetic flux to the rope, cuts Sun during December 2007, named NOAA heliosphere they often exhibit a coherent the tethers of the overlying and constrain- active region 9077. The active region was structure in the solar wind that has an ing field, and aids the acceleration of the studied from its birth on the Sun until it increased magnetic field strength as com- rope away from the Sun. It is this field that produced a CME, allowing the formation of pared to the ambient solar wind flow, an forms the external field of the rope (which the flux rope to be studied in detail. The anomalously low proton and electron tem- is likely to be more twisted than the core) flux rope was built in three phases. Phase perature, and a large magnetic field rotation and which also needs to be known to fore- one is a phase of increasing shear, driven by that is indicative of a particular configura- cast its coupling to the magnetosphere. the dispersal of the surface magnetic field tion containing twisted, or helical, magnetic Therefore, solar observations during the and magnetic reconnection that removes field known as a flux rope (Burlaga et al., dynamic phase of the eruption must also magnetic flux from the surface layer, fol- 1981; Klein and Burlaga, 1982; Cane and be captured. lowing the model of van Ballegooijen and Richardson, 2003). Figure 1 shows a CME Martens (1989). In phase two, there is an whose plasma density distribution strongly Modelling magnetic flux ropes accumulation of a significant amount of flux supports such a twisted configuration. that runs almost parallel to the line that Indeed, it seems that most CMEs may be at the Sun separates the main magnetic polarities of well modelled by a flux rope magnetic con- The starting point for a data-driven simu- the region. In phase three, further magnetic figuration regardless of their source region lation is to model the magnetic flux rope reconnection produces field lines that are (Jian et al., 2006). This has motivated a before its eruption as a CME. The solar twisted around the axial flux (which look search for flux ropes in the solar atmos- observations can be used to determine S-shaped) and which contribute poloidal phere, prior to their eruption as a CME. If the spatial extent of the flux rope, the pre- flux and define the presence of a flux rope. found, the global aspects of their magnetic eruptive orientation of the flux rope axis Images in Figure 2 show these three phases field can be studied to obtain an indica- and even an estimation of the axial flux con- for this solar active region and three others tion of the CME’s geo-effectiveness from tent in special cases (Green et al., 2011). The for emphasis. The observational approach the orientation of the flux rope axis and S-shaped nature of the magnetic configura- to studying the magnetic flux content of the organisation of the twist, leading to a tion is highly suggestive that the flux ropes the rope was brought together with the forecast of the presence of a southward- are weakly twisted structures containing, in flux rope insertion method in Savcheva directed (BZ) magnetic field. We note that most cases, little more than one turn in the et al. (2012), who suggested the reliabil- CMEs may develop sub-structure en route to magnetic field vector. ity of both modelling and observational the Earth (Steed et al., 2011), but this level of The solar data must then be used as the approaches. Gibb et al. (2014) also studied complexity is beyond the scope of current boundary condition for modelling the pre- this region and managed to reproduce the forecasting activity. eruption flux ropes. This has already been main features of the magnetic field evolu- There is increasing evidence that mag- investigated using three approaches. First, tion including the formation of the flux rope netic flux ropes are indeed present in the by the flux rope insertion method (van in the correct location. Figure 3 shows the solar atmosphere before CME onset, per- Ballegooijen, 2004). This method inserts a results of these three studies and directly haps unsurprisingly in the S-shaped (sig- flux rope into a magnetic field configura- compares the observed and modelled mag- moidal) regions that are highly eruptive tion that is extrapolated from solar obser- netic configurations. (Green and Kliem, 2009; 2014; Liu et al., 2010; vations. The location of the rope is guided Green et al., 2011; Zharkov et al., 2011). The by the observation of a filament – a dark three-dimensional structure of the coronal and sinuous feature in the solar atmos- The future field is not directly measurable, but the flux phere. A match between the modelled This paper gives a brief summary of the cur- rope configuration can be inferred from magnetic structure and observed structure rent status of our understanding of coronal 33 mass ejections and the importance of being able to model their magnetic configuration in order to predict severe space weather. We must develop a capability to forecast the strength of the southward magnetic field

component (BZ) of Earth-directed CMEs as well as their velocity and plasma parameters. This is achievable in the near future if the developments we have made in understanding the magnetic configuration of the pre-eruption structure from solar observations, the advances in modelling the mag- Coronal mass ejections netic configuration before the eruption, and data-driven (meaning realistic) simulations of the eruption itself are brought together. From these areas of competency we have the opportunity to work towards two data products that would be valuable for space weather forecasting in relation to CMEs: 1. Exploitation of magnetic field extrapolations and observations of the solar atmosphere to create a near-real-time map of potential space weather ‘hot

Weather Weather – January 2015, Vol. 70, No. 1 spots’ on the Sun. That is, where a magnetic configuration is seen to be forming which has a high likelihood of producing a CME with a strong and sustained

BZ component. 2. Data-driven simulations of CMEs to pro- duce a ‘real Sun’ scenario. These simulations in turn need to be coupled to heliospheric simulations that then prop- agate the CME configuration to 1AU.

Figure 2. Hinode/XRT do X-ray images showing examples of S-shaped (sigmoidal) regions during Acknowledgements the three phases of their evolution. Top row: NOAA region 10930. Second row from the top: NOAA region 8005. Second row from the bottom: an un-numbered region observed on the disc during LMG is grateful to the Royal Society for the February 2007. Bottom row: NOAA region 10977. (Image modified from Green and Kliem (2014).) award of a University Research Fellowship. DB acknowledges support by STFC Consolidated Grant ST/H00260/1. The authors thank Gordon Gibb for supplying a magnetic field extrapolation for Figure 3.

References Allen J, Frank L, Sauer H et al. 1989. Effects of the March 1989 solar activity. EOS Tran s. AGU 70: 1479. Amari T, Aly J-J, Mikic Z et al. 2010. Coronal mass ejection initiation: on the nature of the flux cancellation model. Astrophys. J. 717: L26. Aulanier G, Török T, Démoulin P et al. 2010. Formation of Torus-unstable flux ropes and electric currents in erupting sigmoids. Astrophys. J. 708: 314. van Ballegooi jen AA. 2004. Observations and modeling of a filament on the Sun. Astrophys. J. 612: 519. van Ballegooi jen AA, Martens PCH. 1989. Formation and eruption of solar prominences. Astrophys. J. 343: 971. Figure 3. Comparison between observed and modelled magnetic configuration of solar active Benz AO. 2008. Flare observations. Living region 9077 at two different times. Columns: (a) soft X-ray emission structure observed on the Sun Rev. Sol. Phys. 5: 1. as shown in Green et al. (2011), (b) modelled magnetic configuration of Savcheva et al. (2012), Bobra MG, van Ballegooijen AA, DeLuca (c) modelled magnetic configuration of Gibb et al. (2014). The observations and models suggest EE. 2008. Modeling nonpotential magnetic fields in solar active regions. Astrophys. that a magnetic flux rope has formed by 6 December 2007 around 1550 UTC (bottom row). Note: J. 672: 1209. 34 the size scale varies between panels in the figure. Boteler DH. 2 006. The super storms of Green LM, Kliem B, Wa llace AJ. 2011. structure during sunspot maximum. Space August/September 1859 and their effects Photospheric flux cancellation and associ- Sci. Rev. 33: 219. on the telegraph system. Adv. Space Res. ated flux rope formation and eruption. Siscoe GL, Crooker NU, Ell iott HA. 38: 159. Astron. Astrophys. 526: A2. 2006. Initial-condition influences on CME Burlaga L, Si ttler E, Mariani F et al. 1981. Hood AW, Priest ER. 1981. Solar flares: expansion and propagation. Sol. Phys. Magnetic loop behind an interplanetary Magnetohydrodynamic instabilities, in 239: 293. shock – Voyager, Helios, and IMP 8 obser- Proceedings of the NATO Advanced Study Steed K, Owen CJ, Démoulin P et al. ejections mass Coronal vations. J. Geophys. Res., Space Phys. 86: Institutes Series, Volume 68, Bonas, France, 2011. Investigating the observational 6673. August 25–September 5, 1980. Bonnet RM, signatures of magnetic cloud substruc- Cane HV, Rich ardson IG. 2003. Dupree AK (eds). Springer Netherlands: ture. J. Geophys. Res., Space Phys. 116: Interplanetary coronal mass ejections Dordrecht, Netherlands, pp 509–531. 1106. in the near-Earth solar wind during Illing RME, Hundhause n AJ. 1986. Su Y, van Ballegooijen A, Lites BW et al. 1996–2002. J. Geophys. Res., Space Phys. Disruption of a coronal streamer by an 2009. Observations and nonlinear force- 108: 1156. eruptive prominence and coronal mass free field modeling of active region 10953. Canfield RC, Hudson HS, McKenzie DE. ejection. J. Geophys. Res., Space Phys. 91: Astrophys. J. 691: 105. 10951. 1999. Sigmoidal morphology and eruptive Svalgaard L. 1977. Geomagnetic act ivity: 1 No. 70, Vol. 2015, January – Weather solar activity. Geophys. Res. Lett. 26: 627. Jian L, Russell CT, L uhmann JG et al. dependence on solar wind parameters, Cargill PJ, S chmidt JM. 2002. Modelling 2006. Properties of interplanetary coronal in Coronal Holes and High Speed Wind interplanetary CMEs using magnetohy- mass ejections at one AU during 1995– Streams. Zirker JB (ed.). Colorado Assoc. drodynamic simulations. Ann. Geophys. 2004. Sol. Phys. 239: 393. University Press: Boulder, CO, pp 371–441. 20: 879. Klein LW, Burlaga LF. 1982. Interplanetary Tousey R. 1973. The solar corona, in Chen PF. 2011 . Coronal mass ejections: magnetic clouds at 1AU. J. Geophys. Res., Space Research Conference. Rycroft MJ, models and their observational basis. Space Phys. 87: 613. Runcorn SK (eds). Akademie-Verlag: Berlin, Living Rev. Sol. Phys. 8: 1. Kliem B, Török T. 2006. Torus instability. pp 713–730. Démoulin P, Aulanier G. 2010. Criteria for Phys. Rev. Lett. 96: 255002. Tsurutani BT, Gonzalez WD, Lakhina GS flux rope eruption: non-equilibrium versus Lindsay GM, Luhmann JG, Russell CT et al. 2003. The extreme magnetic storm torus instability. Astrophys. J. 718: 1388. et al. 1999. Relationships between coronal of 1–2 September 1859. J. Geophys. Res. Dungey JW. 1961. Inte rplanetary mag- mass ejection speeds from coronagraph 108: 1268. netic field and the auroral zones. Phys. Rev. images and interplanetary characteristics Valori G, Green LM, Démoulin P et al. Lett. 6: 47. of associated interplanetary coronal mass 2012. Nonlinear force-free extrapolation ejections. J. Geophys. Res., Space Phys. 104: of emerging flux with a global twist Feynman J, Hundhausen AJ. 1994. 12515. Coronal mass ejections and major solar and serpentine fine structures. Sol. Phys. flares: the great active center of March Liu R, Liu C, Wang S et al. 2010. Sigmoid- 278: 73. 1989. J. Geophys. Res., Space Phys. 99: 8451. to-flux-rope transition leading to a loop- Watari S, Detman T. 1998. In situ local like coronal mass ejection. Astrophys. J. shock speed and transit shock speed. Ann. Forbes TG, Isenberg P A. 1991. A catas- Lett. 725: L84. trophe mechanism for coronal mass ejec- Geophys. 16: 370. tions. Astrophys. J. 373: 294. Low BC. 1996. Solar activi ty and the Webb DF, Howard TA. 2012. Coronal corona. Sol. Phys. 167: 217. Forbes TG, Linker JA, Chen J et al. 2006. mass ejections: observations. Living Rev. CME theory and models. Space Sci. Rev. Mackay DH, van Ballegooije n AA. Sol. Phys 9: 3. 123: 251. 2006. Models of the large-scale corona. I. Woo R, Armstrong JW, Sheeley NR et al. Formation, evolution, and liftoff of mag- 1985. Doppler scintillation observations Gibb GPS, Mackay DH, Green LM et al. netic flux ropes. Astrophys. J. 641: 577. 2014. Simulating the formation of a sig- of interplanetary shocks within 0.3AU. J. moidal flux rope in AR10977 from SOHO/ Mackay DH, Green LM, van B allegooijen Geophys. Res., Space Phys. 90: 154. MDI magnetograms. Astrophys. J. 782: 71. A. 2011. Modeling the dispersal of an Xie H, Gopalswamy N, St. Cyr OC et al. active region: quantifying energy input 2008. Effects of solar wind dynamic Gonzalez WD, Joselyn JA, Kamide Y into the corona. Astrophys. J. 729: 97. et al. 1994. What is a geomagnetic storm? pressure and preconditioning on large J. Geophys. Res., Space Phys. 99: 5771. Martin SF. 1980. Preflare conditions, geomagnetic storms. Geophys. Res. Lett. changes and events. Sol. Phys. 68: 217. 35: 6. Gonzalez WD, Echer E, Tsurutani BT et al. 2011. Interplanetary origin of Owens M, Cargill P. 2004. Predictions of Zharkov S, Green LM, Matthews SA intense, superintense and extreme geo- the arrival time of coronal mass ejections et al. 2 011. February 15: Sunquakes pro- magnetic storms. Space Sci. Rev. 158: 69. at 1AU: an analysis of the causes of errors. duced by flux rope eruption. Astrophys. J. Ann. Geophys. 22: 661. 741: 35. Gopalswamy N, Lara A, Lepping RP et al. 2000. Interpla netary acceleration of Pulkkinen T. 2007. Space w eather: terres- coronal mass ejections. Geophys. Res. Lett. trial perspective. Living Rev. Sol. Phys. 4: 1. 27: 145. Russell CT, McPherron RL, Burton RK. Correspondence to: Lucie Green Gosling JT. 1993. The solar flare myth. J. 1974. On the cause of geomagnetic Geophys. Res., Space Phys. 98: 18937. storms. J. Geophys. Res., Space Phys. 79: lucie. gr [email protected] 1105. Green LM, Kliem B. 20 09. Flux rope for- Savcheva A, van Ballegooij en AA. 2009. © 2015 The Authors. Weather published by mation preceding coronal mass ejection John Wiley & Sons Ltd on behalf of Royal onset. Astrophys. J. Lett. 700: L83. Nonlinear force-free modeling of a long- lasting coronal sigmoid. Astrophys. J. 703: Meteorological Society. Green LM, Kliem B. 2014. Observations 1766. of flux rope formation prior to coronal This is an open access article under the mass ejections, in Nature of Prominences Savcheva AS, Green LM, van terms of the Creative Commons Attribution- and their Role in Space Weather Proceedings Ballegooijen AA et al. 2012. Photospheric NonCommercial License, which permits of the International Astronomical Union, flux cancellation and the build-up of sigmoidal flux ropes on the Sun. Astrophys. J. use, distribution and reproduction in IAU Symposium Volume 300. Schmieder B, any medium, provided the original work Malherbe J-M, Wu ST (eds). pp 209–214. 759: 105. Cambridge University Press: Cambridge, Sheeley NR, Howard RA, Koo men MJ is properly cited and is not used for UK and New York, NY. et al. 1982. Observations of coronal commercial purposes. doi:10.1002/wea.2437