An Early Diagnostics of the Geoeffectiveness of Solar Eruptions from Photospheric Magnetic Flux Observations: The Transition from SOHO to SDO
aa r X i v : . [ a s t r o - ph . S R ] F e b Solar PhysicsDOI: 10.1007/ ••••• - ••• - ••• - •••• - • An Early Diagnostics of the Geoeffectiveness of SolarEruptions from Photospheric Magnetic FluxObservations: The Transition from SOHO to SDO
I.M. Chertok · V.V. Grechnev · A.A. Abunin Received ; acceptedc (cid:13)
Springer ••••
Abstract
In our previous articles (Chertok e t al.: 2013, Solar Phys . , 175,and 2015, Solar Phys . , 627), we presented a preliminary tool for the earlydiagnostics of the geoeffectiveness of solar eruptions based on the estimate of thetotal unsigned line-of-sight photospheric magnetic flux in accompanying extremeultraviolet (EUV) arcades and dimmings. This tool was based on the analysisof eruptions observed during 1996 – 2005 with the Extreme-ultraviolet ImagingTelescope (EIT) and the
Michelson Doppler Imager (MDI) on board the
Solarand Heliospheric Observatory (SOHO). Empirical relationships were obtainedto estimate the probable importance of upcoming space weather disturbancescaused by an eruption, which just occurred, without data on the associatedcoronal mass ejections. In particular, it was possible to estimate the intensityof a non-recurrent geomagnetic storm (GMS) and Forbush decrease (FD), aswell as their onset and peak times. After 2010 – 2011, data on solar eruptionsare obtained with the
Atmospheric Imaging Assembly (AIA) and the
Helio-seismic and Magnetic Imager (HMI) onboard the
Solar Dynamics Observatory (SDO). We use relatively short intervals of overlapping EIT – AIA and MDI –HMI detailed observations and, additionally, a number of large eruptions overthe next five years with the 12-hour cadence EIT images to adapt the SOHOdiagnostic tool to SDO data. We show that the adopted brightness thresholdsselect from the EIT 195 ˚A and AIA 193 ˚A image practically the same areas ofarcades and dimmings with a cross-calibration factor of 3.6 – 5.8 (5.0 – 8.2) for theAIA exposure time of 2.0 s (2.9 s). We also find that for the same photosphericareas, the MDI line-of-sight magnetic flux systematically exceeds the HMI fluxby a factor of 1.4. Based on these results, the empirical diagnostic relationshipsobtained from SOHO data are adjusted to SDO instruments. Examples of a post-diagnostics based on SDO data are presented. As before, the tool is applicable Pushkov Institute of Terrestrial Magnetism, Ionosphere andRadio Wave Propagation (IZMIRAN), Troitsk, Moscow,108840 Russia, email: [email protected] ; [email protected] Institute of Solar-Terrestrial Physics SB RAS, LermontovSt. 126A, Irkutsk 664033, Russia, email: [email protected]
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 1 hertok, Grechnev, and Abunin to non-recurrent GMSs and FDs caused by nearly central eruptions from activeregions, provided that the southern component of the interplanetary magneticfield near the Earth is predominantly negative, which is not predicted by thistool.
Keywords:
Solar eruptions; Coronal mass ejections; Dimmings; Arcades; Mag-netic flux; Forbush decreases; Geomagnetic storms
1. Introduction
Coronal mass ejections (CMEs) and their interplanetary counterparts, interplan-etary coronal mass ejections (ICMEs), and in particular magnetic clouds, areprime drivers of the most severe non-recurrent space weather disturbances. Mostsignificant among them are major geomagnetic storms (GMSs) ( e.g.
Gosling,1993; Bothmer and Zhukov, 2007; Gopalswamy, Tsurutani, and Yan, 2015) andForbush decreases (FDs) of the intensity of galactic cosmic rays (Cane, 2000;Belov, 2009; Richardson and Cane, 2011). One of the most important challengesof solar-terrestrial physics and space weather prediction is the diagnostics of thegeoefficiency of CMEs, i.e. an approximate estimation and forecast of possiblenon-recurrent GMS and FD parameters from observed characteristics of an erup-tion that has just occurred. On the Sun, CME eruptions are accompanied bysuch phenomena as bright post-eruption arcades (Kahler, 1977; Sterling et al. ,2000; Hudson and Cliver, 2001; Tripathi, Bothmer, and Cremades, 2004; Yashiro et al. , 2013) and large-scale dark dimmings (Thompson et al. , 1998; Hudson andCliver, 2001; Harra et al. , 2011). They are observed particularly in the extreme-ultraviolet (EUV) range and represent the structures and areas involved in theCME process.Our previous articles (Chertok et al. , 2013, 2015 hereafter referred to as Arti-cle I and Article II) showed the total unsigned magnetic flux of the longitudinalfield at the photospheric level within the arcade and dimming areas to be asuitable quantitative parameter for the earliest diagnostics of the geoefficiencyof solar eruptions. This approach is based on widely accepted concepts relatingpaired core dimmings to the footpoints of an erupting CME flux rope and thepost-eruption arcade to the magnetic structures remaining after reconnectionthat formed this flux rope. We studied events of Solar Cycle 23 during 1996 –2005 in which sources of major non-recurrent GMSs with a geomagnetic indexDst < −
100 nT were reliably identified as near disk-center active regions (ARs).These eruptions were analyzed using data from the
Solar and HeliosphericObservatory (SOHO: Domingo, Fleck, and Poland, 1995), namely solar imagesobtained with the
Extreme-ultraviolet Imaging Telescope (EIT: Delaboudini`ere etal. , 1995) in the 195 ˚A channel and magnetograms acquired with the
MichelsonDoppler Imager (MDI: Scherrer et al. , 1995). As a result, clear correlationswere found between the erupted magnetic flux under the arcades and dimmingsfollowing eruptions from ARs and the amplitude of GMSs (Dst and Ap indexes)and FDs, as well as their temporal parameters (intervals between the solareruptions and the GMS onset and peak times). The larger the erupted flux,
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 2 iagnostics of the Geoeffectiveness of Solar Eruptions the stronger the GMS or FD intensities are and the shorter the ICME transittime is. These correlations indicate that the quantitative characteristics of majornon-recurrent space weather disturbances are largely determined by measurableparameters of solar eruptions, in particular by the magnetic flux within thearcade and dimming areas, and can be tentatively estimated in advance with alead time from one to four days.These dependencies expressed in corresponding empirical relationships con-stitute a preliminary tool based on SOHO data for an early diagnostics ofgeoefficiency of solar eruptions and a short-term forecasting of the main param-eters of non-recurrent space weather disturbances. However, at the end of 2010July, the synoptic 12-min cadence SOHO/EIT observations in the 195 ˚A channelwere replaced by obtaining a couple of images per day only, at around 01:13 and13:13 UT (see the EIT catalog at http://umbra.nascom.nasa.gov/eit/eit-catalog.html ).In addition, the solar magnetic field observations with the SOHO/MDI magneto-graph were terminated in 2011 April (see the MDI Daily Magnetic Field SynopticData at http://soi.stanford.edu/magnetic/index5.html ).Solar activity in the present Cycle 24 is relatively low. This has resulted ina smaller number of non-recurrent geospace disturbances initiated by eruptionsfrom ARs that were relatively weak (Gopalswamy, Tsurutani, and Yan, 2015).For this reason, it is not possible to repeat the analysis of Articles I and IIfor the data of the
Solar Dynamic Observatory (SDO: Pesnell, Thompson, andChamberlin, 2012), which provides regular high-quality observations starting inearly 2010 May. To use our tool at the present time for the early diagnosis ofsolar eruptions, our procedures should be upgraded for the usage of SDO data.To shift from SOHO data to SDO correctly, we examine our diagnostic tool usingboth SOHO and SDO observations in the intervals when they overlap. For theextraction of the arcade and dimming areas it is reasonable to use now the SDO
Atmospheric Imaging Assembly (AIA: Lemen et al. , 2012) images produced inthe 193 ˚A channel instead of the SOHO/EIT 195 ˚A images, because both EUVchannels have a close temperature response with peaks at about 1.3 – 1.5 MK.Calculation of the magnetic flux in the extracted areas should be done with datafrom the SDO
Helioseismic and Magnetic Imager (HMI: Scherrer et al. , 2012).These issues are the subject of the present article.
2. Data and Methodical Issues
Our approach invokes a widely accepted view on the flux rope formation insolar eruptions due to reconnection during flares. Conjugate footpoints of anerupted flux rope are revealed by paired core dimmings (Hudson and Cliver,2001). Observational studies confirm this view (see, e.g. , Qiu et al. , 2007; Mik-lenic, Veronig, and Vrˇsnak, 2009). In particular, Qiu et al. (2007) establishedfor several events a quantitative correspondence between the flare-reconnectedmagnetic flux with the poloidal (azimuthal) flux in the corresponding magneticclouds near Earth. A smaller toroidal (axial) flux corresponded to the magneticflux in dimmed regions.
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 3 hertok, Grechnev, and Abunin
Any erupting structure initially has magnetically conjugate bases, even if iteventually becomes disconnected from the Sun. The magnetic flux in one baseshould be the same through the erupting structure and in the conjugated base, i.e. the positive and negative magnetic fluxes should be exactly balanced. Acorresponding photospheric magnetogram presents, along with the bases of theerupting structure, numerous footpoints of compact loops, which are not involvedin the eruption. The magnetic flux computed from a photospheric magnetogramshould therefore be basically excessive. To approach a real erupted magneticflux, magnetic fields should be extrapolated into the corona. Qiu et al. (2007)extrapolated magnetic fields to a fixed height of 2 Mm and calculated signed fluxin opposite-polarity regions separately. The flux balance defined as a ratio of thepositive to negative flux ranged typically between 0.8 and 1.5. In a case studyof Uralov et al. (2014), the flux balance was reached by adjusting the height ofthe extrapolation, and the ultimate result corresponded well to the estimationsfor the near-Earth magnetic cloud.The last way (possibly, in an elaborated form) appears to be the most promis-ing for accurate estimations of the erupted magnetic flux. However, this laboriousway is difficult to use in statistical studies such as Article I presented. To expeditecalculations, in this article, we measure the magnetic flux directly from photo-spheric magnetograms, without extrapolation. The balanced conjugate bases oferupting structures are not known in this situation and we use unsigned magneticfluxes under arcades and dimming regions for simplicity and consistency withArticles I and II. Such measurements overestimate a real erupted flux by a factorof two, due to the summation of the outgoing and incoming fluxes. Our directmeasurement of the magnetic flux from the photospheric magnetograms withoutextrapolation causes an additional overestimate by a factor varying from oneevent to another. Thus, the actual flux in a magnetic cloud should presumablybe less than our estimates by a factor of 4 – 10. A constant factor does not affectthe statistical patterns we discuss, and an unknown variable factor contributesto the scatter, along with other circumstances, which we do not consider, such asthe sign of the southern component of the interplanetary magnetic field, Bz , thepresence of a possible negative Bz in the leading or trailing part of a magneticcloud, and others.We extract arcades and dimmings in EIT and AIA images with the sametechniques as those used in Article I, based on formal criteria referring to abrightness analysis, as described below. These criteria detect the features ofinterest more or less reliably; nevertheless, a manual assistance in the interac-tive mode is generally required to include separate disconnected regions or toeliminate irrelevant ones in full-disk images.To facilitate interactive manipulations, we reduce both EIT and AIA imagesby rebinning them to a common format of 512 ×
512 pixels, as we previously han-dled SOHO data (Article I). Most EIT images and all MDI magnetograms have1024 × ×
512 pixels; then, these areunchanged. SDO/AIA and HMI images have 4096 × SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 4 iagnostics of the Geoeffectiveness of Solar Eruptions data, an advantage of reduced images is a slightly enhanced signal-to-noise ratioand a smaller contribution from compact defects.The pixel size of the full-resolution SOHO/EIT images is 2 . ′′ . SOHO islocated at the L1 Lagrangian point 0.01 AU sunward from the Earth. The EITpixel size converted to the near-Earth vantage point of SDO is 2 . ′′ × .
99 =2 . ′′ . The ratio of the linear scales in the EIT and AIA (0 . ′′ pixel size) imagesreduced to the 512 ×
512 pixels format is (2 . ′′ × / (0 . ′′ ×
8) = 1 . . ′′ and HMI magnetograms to match the reduced AIA imageswith a pixel size of 4 . ′′ . We refer henceforth to the reduced images (512 × http://umbra.nascom.nasa.gov/eit/eit-catalog.html and http://jsoc2.stanford.edu/data/aia/synoptic/ ). EIT images were processed bythe standard EIT PREP routine and we handled level 1.5 AIA images withoutany additional pre-processing. We only normalized all AIA images observedduring each event to a common exposure time, given by a pre-event image.The solar rotation in the analyzed images was removed and a background pre-eruption image was subtracted from each subsequent one to obtain fixed-basedifference images.For arcades, a criterion turned out to be appropriate, which extracted thearea around the flare site, where the brightness exceeded 5 % of the maximumover the Sun during the event (in an image with the brightest flare emission).This criterion based on a relative brightness threshold has been used for bothEIT and AIA images without any adjustment.The detection of dimming regions is more complex. Parameters of dimmingswere computed from the so-called “portrait”, which shows in a single compositeimage all dimmings appearing during the event. The “dimming portrait” isgenerated by finding a minimum brightness in each pixel over the entire fixed-base difference set (see Chertok and Grechnev, 2005). Reinard and Biesecker(2008, 2009) concluded that dimmings only appeared in events associated withfast CMEs and strong flares. Indeed, thresholding of difference images by acertain value (say, −
50 for EIT images) reliably detects dimmings in flare-relatederuptions from active regions, where pre-eruptive structures are bright, but canbe insufficient in non-flare-related events outside of active regions, where thebrightness of pre-eruptive structures is modest. To detect dimmings in theselatter events, where depressions are generally shallower, the relative brightnessthresholding is more efficient. A brightness depression deeper than −
40% of apre-event level is an optimal criterion for extraction of significant core dimmingslocated near the eruption center and obviously related to the eruption. On theother hand, this criterion is too strong for some flare-related eruptions. Wetherefore had to use a combined dimming-detecting criterion based on bothrelative and absolute thresholds; the latter are not the same for EIT and AIAdue to their different sensitivity.
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 5 hertok, Grechnev, and Abunin
The difference in the sensitivity of the EIT and AIA telescopes can be com-pensated, if a ratio is known of the AIA to EIT responses, when both telescopesobserve the same structure. Counts per pixel in an AIA image divided by thisratio (cross-calibration factor, CCF) should become close to those in a corre-sponding EIT image, and both relative and absolute thresholds can be used forany image independent of the instrument that produced it.We calculated the CCF for each event as a ratio of the AIA and EIT differencesbetween the brightness of the quiet Sun and that of the sky, i.e.
CCF = ( B qs − B sky ) AIA / ( B qs − B sky ) EIT . These levels were evaluated as the positions of thepeaks of the histograms representing the brightness distributions (number ofpixels vs. brightness represented by the counts per pixel) in the EIT and AIAimages within a disk of 0 . R ⊙ (quiet Sun) and outside 1 . R ⊙ (sky) from theSun center. The peaks are centered at the highest-probability values, aroundwhich pixels with a highest occurrence frequency are concentrated.The pixels corresponding to bright structures fall in the quiet Sun histogramconsiderably to the right from the peak and those corresponding to dark coronalholes fall left from the peak. Statistical contributions of the bright and darkstructures determined by their relative areas with respect to the solar disk arerelatively small and they do not displace the peak of the histogram. Similarly,bright off-limb structures do not affect the position of the peak correspondingto the sky level. This reliable technique has been widely used for calibrationof microwave images produced by radio heliographs ( e.g. Hanaoka et al. , 1994;Grechnev et al. , 2003; Kochanov et al. , 2013).The brightness of the sky, B sky , is close to zero in both EIT images pre-processed with the EIT PREP routine and level 1.5 AIA images. Thus, dividingan AIA image by the CCF brings it to the EIT data range.
3. Comparison of Areas Involved in Eruptions
Firstly, it is necessary to compare the configurations and areas of dimmings andarcades extracted for the same events in the SOHO/EIT 195 ˚A and SDO/AIA193 ˚A images. Two groups of eruptions from the central part of the solar disk(mainly within ± ◦ , sometimes ± ◦ from the disk center) observed simultane-ously with SOHO and SDO are considered (Table 1). Group A includes relativelyweak eruptions in 2010 May – July for which EIT images observed with a detailed12-min interval were still available. Because of the low solar activity at that time,the events were selected by inspecting daily EIT and AIA movies showing signsof an eruption regardless of the importance and nature of the accompanying softX-ray flares. A number of them were classified as filament eruptions outside ARs,although ultimately we are interested in eruptions from ARs. In contrast, GroupB consists of evident, strong eruptions associated with flares of importance largerthan M1.0 that occurred during 2011 – 2015. For these events we are forced to usethe EIT images available with a 12-hour interval and co-temporal AIA images.Considering lifetimes of arcades and dimmings of several hours, an additionalrequirement for these events was that the GOES peak of the corresponding X-ray flares should occur not earlier than three hours before the EIT observational SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 6 iagnostics of the Geoeffectiveness of Solar Eruptions time, which is usually at 01:13 and 13:13 UT. An event was also required tobe isolated in a sense that in the considered 12-hour interval between two EITimages there were no other events of a comparable intensity.In most cases of Group A, an interval starting short before the eruption onsetand lasting 3 – 4 hours afterwards was considered, i.e. a set of 15 – 20 images wasanalyzed. During this interval, the main arcades and dimmings were alreadyfully formed. For events of Group B, the first image was subtracted from thesecond one to reveal the arcades and dimmings in the difference image. In someevents of Group B, the time of the second image was close to the time of themaximum of the corresponding soft X-ray flare.As it is known, the area of a post-eruption arcade increases with time. There-fore, to avoid ambiguity, extraction of an arcade area in events of Group A isperformed in an image temporally close to the maximum of the EUV flux fromthe selected area. Usually this time is close to the peak time of a correspondingGOES soft X-ray flare or somewhat later. It is clear that for Group B events theextraction time was defined by the timing of the second EIT image and mostoften it did not correspond to the maximum dimming and arcade areas for thegiven event.Table 1 lists eruptions of Groups A and B selected for comparison of thearcade and dimming areas in the EIT 195 ˚A and AIA 193 ˚A images. Each eventis specified by its number with an index A or B in the first column, date inthe second column, time in the third column of an image used for the arcademeasurement for Group A and the soft X-ray flare peak time for Group B, GOESclass of the related flare in the fourth column, and an approximate position of aneruption or flare in the fifth column. We do not present the exposure times ( τ exp )of the analyzed EIT images that are given in the headers of the correspondingFITS files, because almost all EIT images (except those for two events) had τ exp ≈ . τ exp ≈ . τ exp , although some imagesof Group B events corresponding to the flare peaks were produced with a muchshorter exposure time of up to τ exp ≈ .
07 s.The results of calculating the CCF for all events and for two AIA exposuretimes typically used without flares are presented in the sixth column of Table 1(separated by a slash for τ exp = 2 . τ exp = 2 . τ exp ≈ . τ exp ≈ . × × i.e. CCF( τ exp = 2 . / CCF( τ exp = 2 .
0) should be equal to 2 . / . . . ± . SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 7 hertok, Grechnev, and Abunin
Table 1.
Analyzed eruptions, a cross-calibration factor (CCF) between the EIT and AIAimages and areas of the extracted dimmings (Dim) and arcades (Arc).No Date Time GOES Position CCF SOHO/EIT SDO/AIAUT class for τ exp areas [pixels]2.0/2.9 s Dim Arc Dim Arc1A 2010-05-23 19:35 B1.3 N18W15 4.8/6.8 1828 761 1981 6902A 2010-05-24 15:48 B1.1 N18W27 5.1/7.3 1131 306 1222 3843A 2010-05-31 22:24 A6.5 N25W27 4.6/6.6 1095 414 1303 3634A 2010-06-07 19:48 B2.0 N22E26 4.7/6.8 119 75 113 485A 2010-06-12 01:14 M2.0 N23W43 5.4/7.7 205 19 218 456A 2010-06-17 11:12 B5.0 N28E41 4.9/7.2 48 210 51 2187A 2010-06-29 13:47 A5.0 S20W22 1.0/1.4 413 375 625 4778A 2010-06-29 16:23 B1.3 N17W20 0.9/1.3 69 94 65 689A 2010-07-14 12:48 C1.4 N21E06 5.3/7.6 0 419 0 32510A 2010-07-16 15:48 B1.3 S21W22 5.8/8.2 922 49 875 4411A 2010-07-17 18:23 C2.4 N20W33 4.6/7.1 1496 97 1419 10912A 2010-07-18 07:12 B1.0 N30E20 5.4/7.8 206 147 242 33613A 2010-07-19 09:23 B4.0 N30W05 5.1/7.4 1316 267 1332 26214B 2011-09-06 22:20 X2.1 N14W18 5.2/7.5 1395 276 1140 28915B 2012-06-13 13:17 M1.2 S16E18 5.0/7.3 1054 253 984 20316B 2012-08-11 12:20 M1.0 S28W38 4.5/6.5 544 487 639 45117B 2013-08-12 10:41 M1.5 S17E19 4.3/6.2 249 118 242 18918B 2013-10-13 00:43 M1.7 S22E17 5.1/7.3 1126 111 1170 7919B 2013-10-24 00:30 M9.3 S10E08 4.9/7.1 1783 80 1820 6220B 2014-01-07 10:13 M7.2 S13E11 3.8/5.5 0 765 0 82121B 2014-02-04 01:23 M3.8 N09W13 3.6/5.1 50 116 54 7922B 2014-04-18 13:03 M7.3 S18W37 4.0/5.8 303 241 600 17723B 2014-10-24 21:41 X3.1 S16W21 4.5/6.5 0 733 45 74224B 2015-03-11 00:02 M2.9 S16E28 3.7/5.3 321 105 302 12425B 2015-03-15 23:22 M1.2 S17W35 3.8/5.4 273 402 306 33626B 2015-03-16 10:58 M1.6 S17W39 4.8/6.9 134 468 149 56427B 2015-06-21 01:42 M2.0 N12E13 3.8/5.5 238 342 263 24228B 2015-11-09 13:12 M3.9 S11E41 3.6/5.0 137 131 321 83 For each event we extracted the arcades and dimmings in the images producedwith both EIT and AIA and compared their characteristics, including their quan-titative areas. In the case of the AIA images, we made it for two combinations ofthe exposure times (2.0 and 2.9 s) and corresponding CCF. As noted in Section 2,in the images reduced to the same common format of 512 ×
512 pixels, the AIApixel size exceeds the corresponding EIT parameter by a factor of 1.085. In thisstudy the dimming and arcade areas are expressed in pixels. Consequently, tobring the AIA areas to the EIT scale, the AIA areas were divided by a factor(1 . ≈ . SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 8 iagnostics of the Geoeffectiveness of Solar Eruptions C r o ss - c a li b r a t i on f a c t o r Group A, 2010 Group B, 2011-2015
Event number according to Table 1 τ exp = 2.9 s τ exp = 2.0 s Figure 1.
Cross-calibration factor (CCF) between the EIT 195 ˚A and AIA 193 ˚A images withan AIA exposure time of 2.0 and 2.9 s for Group A and B events labeled by sequential numbersas indicated in Table 1.
AIA dimming and arcade areas brought to the EIT pixel scale and those ex-tracted from the EIT images (seventh and eighth columns) are presented. Thecorresponding scatter plots are shown in Figure 2. The majority of points cor-responding to the EIT and AIA dimmings (Figure 2a) and arcades (Figure 2b)lie near the dotted bisector ( y = x ) line, which also nearly coincides with thebest linear fit with a correlation coefficient of 0.98 for the dimmings and 0.95for the arcades. We pay special attention to two events with a relatively largedifference between the EIT and AIA dimming areas (event 22B) and arcade area(event 12A). Event 12A occurred almost without any enhancement in the softX-ray flux, which stayed at the B1 background level. Accordingly, its arcade wasvery faint and, based on the accepted criteria, was extracted as small separatefragments. In event 22B, the scarcely extracted weak dimmings had a similarfragmentary character.Another important result is that the extracted arcades and dimmings inpractically all analyzed events coincide in the EIT and AIA images both intheir areas and configurations. This is illustrated in Figure 3, where the closecontours of the EIT and AIA dimmings and arcades of two Group A eruptionsare presented by different colors on the background of the SDO/HMI pre-eventmagnetograms (bottom). The AIA difference images are very similar to thoseshown in Figures 3a and 3b.All of these comparisons provide a basis for a general conclusion that withthe adopted criteria, our procedures extract in the SDO/AIA 193 ˚A images thesame dimmings and arcades as in the SOHO/EIT 195 ˚A images and AIA datacan be used instead of EIT. SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 9 hertok, Grechnev, and Abunin S D O / A I A d i mm i ng a r ea [ p i x e l s ] Detailed data of 2010Limited 12 h data of 2011-2015 a y = x S D O / A I A a r c ade a r ea [ p i x e l s ] Detailed data of 2010Limited 12 h data of 2011-2015 b y = x Figure 2. (a) Scatter plots of dimming and (b) arcade areas extracted in the EIT 195 ˚A andAIA 193 ˚A images. The AIA areas are brought to the EIT pixel scale. Open and filled circlesdenote events of Groups A and B, respectively. The dotted lines correspond to the equal areas( y = x ).
4. Comparison of Erupted Magnetic Fluxes
The second procedure, which should be considered, is the comparison of theerupted magnetic fluxes as measured by the SOHO/MDI and SDO/HMI mag-netographs. In the period between the beginning of HMI observations in 2010May and end of MDI observations in 2011 April, there were no large eruptionsin ARs. Moreover, many of the Group A events considered in the precedingsection were associated with filament eruptions outside ARs and had very smallmagnetic fluxes. It is not reasonable to compare the MDI and HMI fluxes ofthese eruptions. Instead, we will compare the total magnetic fluxes of the 34largest ARs, which were observed near the solar disk center during the con-current MDI and HMI observations and had the sunspot area greater than 100
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 10 iagnostics of the Geoeffectiveness of Solar Eruptions a EIT b
EITc HMI d HMI
Figure 3.
Top: bright post-eruption arcades and dark dimmings in SOHO/EIT 195 ˚A differ-ence images for eruptions 1A (2010 May 23, panel a) and 15B (2012 June 13, panel b). Bottom:corresponding SDO/HMI pre-event magnetograms (panels c and d) overlaid with contours ofextracted dimmings (dark green for AIA and light green for EIT) and arcades (blue for AIAand pink for EIT). The contours illustrate the acceptable correspondence between the areasselected from the images produced by the two instruments. millionths of the Sun visible hemisphere ( µ Hem) (see, e.g. , the Solar Monitor site ). Although the measurements with HMI and MDIhave been compared previously, e.g. by Liu et al. (2012), we are not aware of suchcomparisons for magnetic fluxes erupted from large, developed active regions,where magnetic fields are strong and saturation-like distortions are probable.The observation date, time, NOAA number, coordinates, and maximum areasof these ARs are listed in Table 2. For each of these ARs, using data from bothmagnetographs, we calculated the total unsigned magnetic flux from the line-of-sight photospheric field within a single topologically-connected area, wherethe magnetic field strength exceeded 10 % of the maximum value over the entiremagnetogram, but was not less than 15 G. All full-disk magnetograms were
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 11 hertok, Grechnev, and Abunin
Table 2.
The SOHO/MDI and SDO/HMI magnetic fluxes of the largestARs over the period from 2010 May to 2011 April.Date Time AR Position Area Magnetic fluxhh, UT number [ µ Hem] [10 Mx]Φ
MDI Φ HMI reduced to the 512 ×
512 pixels format. For MDI we used level 1.8 magnetogramsrecalibrated in 2008 December. The corresponding MDI and HMI FITS files weredownloaded from the Stanford University sites of the MDI Daily Magnetic FieldSynoptic Data ( http://soi.stanford.edu/magnetic/index5.html ) and from the Stan-ford Joint Science Operations Center ( http://jsoc2.stanford.edu/data/hmi/fits/ ). SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 12 iagnostics of the Geoeffectiveness of Solar Eruptions Mx]0100200300400 M D I f l u x [ M x ] y = . Figure 4.
Relationship between the MDI and HMI magnetic fluxes of the largest ARs overthe period from 2010 May to 2011 April (see Table 2). The dotted line corresponds toΦ
MDI = 1 . HMI . The relationship between the MDI, Φ
MDI , and HMI, Φ
HMI , AR magneticfluxes calculated in this way is shown in Figure 4. Its best linear fit in a widerange of the flux values of Φ
MDI ≃ −
350 and Φ
HMI ≃ −
270 (in 10 Mxunits) is Φ
MDI = (3 . ± .
5) + (1 . ± . HMI with a correlation coefficientof r ≈ .
99. Within the measurement errors, the factor we obtained from theanalysis of the magnetic flux of ARs is consistent with the result of Liu et al. (2012), who established by a pixel-by-pixel comparison that the line-of-sightpixel-averaged magnetic signal inferred from MDI magnetograms was greaterthan that derived from the HMI data by the same scaling factor of 1.4. A closefactor of 1.35 was also found by Svalgaard and Sun (2016). The differencesbetween the measurements from MDI and HMI data can be due to their differentcalibration and a number of other factors (see Riley et al. , 2014; Watson, Penn,and Livingston, 2014; Couvidat et al. , 2016). Thus, in the transition from SOHOto SDO data, the relation Φ
MDI = 1 . HMI should be used.
5. Transition Procedure
The results of the two previous sections allow us to adapt the SOHO diagnostictool presented in Articles I and II to SDO data and current measurements.The updated tool does not need SOHO data. For the extraction of dimmingand arcade areas for a particular eruption, it is sufficient to apply the cross-calibration factor between the EIT 195 ˚A and AIA 193 ˚A images. It is possibleto simply adopt CCF ≈ τ exp ≈ . ≈ τ exp ≈ .
9, based onFigure 1 for the first years. After that, one should download a FITS file of theHMI magnetogram that precedes the onset of an eruption and a number of AIA
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 13 hertok, Grechnev, and Abunin files covering approximately the duration of an associated soft X-ray flare. Byanalogy with the SOHO diagnosis, for the extraction of the SDO dimmings andarcades it is sufficient to take the AIA files with a 12-min interval. All AIA imagesshould be corrected to a single pre-event exposure time (2.0 or 2.9 s) and dividedby the CCF before the extraction of the arcade and dimming areas. Then, bycoaligning the extracted areas with the HMI magnetogram, the correspondingerupted flux, Φ
HMI , is calculated.Possible magnitudes as well as the onset and peak times of a geomagneticstorm and Forbush decrease are estimated by means of conversion of the corre-sponding empirical expressions, obtained in Articles I and II for SOHO data, inaccordance with a relation Φ
MDI = 1 . HMI . For SDO data, these expressionsare as follows (again Φ
HMI is in units of 10 Mx): • GMS intensity (Dst and Ap indexes)Dst [nT] = 30 − . HMI + 3 . / , Ap [2nT] = 1 . HMI . • FD magnitude A F [%] = − . . HMI . • The onset (∆ T o ) and peak (∆ T p ) transit times, i.e. the intervals betweenthe eruption time (maximum time of an associated soft X-ray burst) andthe start and peak of a corresponding GMS∆ T o [h] = 98 / (1 + 0 . HMI ) , ∆ T p [h] = 118 / (1 + 0 . HMI ) .
6. Examples
Now we consider some examples of an SDO-based post-diagnosis related toseveral large eruptions in ARs located not far from the solar disk center andto major GMSs, which occurred during the current Solar Cycle 24. As alreadynoted, due to a relatively low level of solar activity after 2009, only few majorGMSs with Dst < −
100 nT occurred. Most of the GMSs were caused by filamenteruptions outside ARs and sometimes by high-speed solar wind from coronalholes (Gopalswamy et al. , 2015; Gopalswamy, Tsurutani, and Yan, 2015). Forinstance, the strongest geomagnetic storm of the current solar cycle with aminimum Dst = −
228 nT on 2015 March 17 was initiated by a large south-west filament eruption near AR 12297. Note that, according to our estimations,its erupted magnetic flux, Φ
HMI ≈ . × Mx (corresponding to the fluxmeasured by MDI of Φ
MDI ≈ . × Mx), was larger than the fluxes infilament eruptions, which caused the GMS during Solar Cycle 23 (blue trianglesin Figure 4 of Article I).
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 14 iagnostics of the Geoeffectiveness of Solar Eruptions
Table 3 presents the results of the SDO-based diagnosis of five flare-associatederuptions in ARs carried out according to the procedure described in the pre-vious section. The table lists parameters of a solar event in the second tosixth columns, including the date, peak time, coordinates, and GOES classof a related flare (second to fifth columns), and a total unsigned magneticflux calculated from HMI magnetograms in dimmings and arcades in the sixthcolumn. The seventh to nineteenth columns are related to the geospace dis-turbance, including its peak date and time, and estimated (letter code “est”)and observed (“obs”) parameters of a corresponding GMS and FD. Informationon the observed hourly Dst index was taken from the WDC2 Kyoto service( http://wdc.kugi.kyoto-u.ac.jp/dstdir/index.html ), while the values of the three-hour Ap index were estimated in the GeoForschungsZentrum (GFZ), Potsdam( ftp://ftp.gfz-potsdam.de/pub/home/obs/kp-ap/wdc/ ). The onset transit time,∆ T o , is defined as an arrival time of the corresponding interplanetary disturbance(shock wave) at Earth indicated by the geomagnetic storm sudden commence-ment (SSC) ( ). Following Article I, the FD maxi-mum magnitude is adopted; this magnitude corresponds to a cosmic ray rigidityof 10 GV evaluated from data of the world network of neutron monitors using theglobal survey method (Krymskii et al. , 1981; Belov et al. , 2005). Additionally,the eighteenth and nineteenth columns of Table 3 list the hourly strength of thetotal interplanetary magnetic field near the Earth, Bt , and its Bz componentaccording to the Operating Missions as a Node on the Internet (OMNI) data( ftp://spdf.gsfc.nasa.gov/pub/data/omni/low res omni/ ).Table 3 confirms the results of Articles I and II that the early diagnosis of theeruptions, based on the erupted magnetic flux evaluated in this case from SDOdata, provides an approximate assessment for the importance of the related spaceweather disturbances. Particularly, the estimated values of Dst, Ap, ∆ T o , ∆ T p ,and FD are comparable with the observed ones. According to the NOAA spaceweather scale ( ), four events(Nos. 1 – 3 and 5) are classified as G2 – G3 (moderate – strong) storms, and one(No. 4) as G4 (severe) storm, judging from both the estimated and observedGMS intensity.Table 3 presents an apparent scattered correspondence between the estimatedand observed parameters of the GMSs and FDs. Indeed, the larger the eruptedflux, the stronger the actual intensity of the GMSs and FDs, and shorter thetime intervals between the parent eruption and the GMS onset and peak are.For example, the Pearson correlation coefficient between the estimated Dst andits observed value is 0.61. On average, Dst obs / Dst est = 0 . ± .
19 ( ± Bz component seems to be comparable, − Bz/Bt = 0 . ± . ± Bz . There is also a close correspondence with acorrelation coefficient of 0.91 between the erupted flux (sixth column) and thetotal magnetic field strength ( Bt in the eighteenth column) brought to Earth bythe ICMEs.The event on 2016 June 21 (No. 4 in Table 3) with the largest erupted fluxresulted in the most intense GMS and FD, having also the shortest transit timesand strongest interplanetary magnetic field. In some events, for example Nos. 1 SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 15 h e r t o k , G r ec hn e v , a nd A bun i n Table 3.
Results of the SDO-based post-diagnosis of some eruptions of Solar Cycle 24.No Eruption Geomagnetic storm Forbush Magneticdecrease field [nT]Date Time Position Flare Φ
HMI
Peak Dst [nT] Ap [2nT] ∆ T o [h] ∆ T p [h] [%] Bt Bz
UT class [10 Mx] dd/hh est obs est obs est obs est obs est obs1 2012-03-07 00:24 N17E27 X5.4 178.5 09/09 -178 -131 200 132 47 35 59 56 7.2 11.7 23.1 -16.12 2012-07-12 16:49 S15W01 X1.4 197 15/19 -188 -127 220 132 44 49 56 65 8.0 6.4 27.3 -17.73 2013-03-15 06:58 N11E12 M1.1 85.5 17/21 -116 -132 96 111 64 47 80 62 3.3 4.6 17.8 -144 2015-06-21 02:06 N12E13 M2.6 234.4 23/05 -208 -204 263 236 40 40 51 52 9.5 8.4 37.7 -26.35 2015-12-28 12:45 N19W22 M1.8 128.3 01/01 -147 -117 144 80 55 63 69 84 5.1 4.3 The observed FD magnitude was evaluated from data of two high-latitude different-hemisphere stations, Thule and McMurdo.
SOLA:transition_prep.tex;17September2018;7:49;p.16 iagnostics of the Geoeffectiveness of Solar Eruptions and 2, the estimated GMS intensity, measured both by the Dst and Ap indexes,markedly exceeds the observed one, while the magnitudes of the FDs are muchcloser. This is apparently due to the fact that in these cases the negative Bz component, which determines the GMS intensity, but not considered in ourpreliminary tool, amounts to only part of the total interplanetary magneticstrength, which determines the FD magnitude.As for the temporal parameters of GMSs, a close correspondence between theestimated and observed values of both ∆ T o and ∆ T p is present for event No. 4. Inother cases, some differences can be seen. Our simple tentative estimates havebeen done as if they were issued right after an eruption, without taking intoaccount preceding activity, actual magnetic field and plasma distributions in thecorona, ICME drag in the solar wind, and other factors.
7. Concluding Remarks
We compared quantitative parameters of CME-associated post-eruption arcadesand dimmings observed with EUV telescopes and magnetographs aboard theSOHO and SDO spacecraft. Two basic facts have been established. First, withthe adopted thresholds of relative brightness changes, practically the same arcadeand dimming areas are extracted from EIT 195 ˚A and AIA 193 ˚A images, iftheir cross-calibration factor in a range of 3.6 – 5.8 and 5.0 – 8.2 is taken intoaccount for the AIA exposure time 2.0 and 2.9 s, respectively. Second, for thesame photospheric areas of strong magnetic fields in large active regions, theMDI line-of-sight magnetic flux systematically exceeds the HMI flux by a factorof 1.4.These results allowed us to upgrade the tool for the early diagnostics of AReruptions described in Articles I and II for SOHO data to the current SDO ob-servations. Empirical relationships are obtained to connect the erupted magneticflux measured from SDO data with possible intensity and temporal parametersof forthcoming non-recurrent GMSs and FDs. The case studies presented hereconfirm that the updated diagnostic tool based on SDO data also producesacceptable results, providing a prompt and sufficiently correct assessment of theimportance of forthcoming space weather disturbances using only magnetic fluxwithin the arcade and dimming areas. The tool presented here and previously inArticles I and II corroborates the idea that parameters of solar eruptions, CMEsand ICMEs, and geospace disturbances are largely determined not only by thecharacteristics of associated ARs and flares, but by a measurable quantity asthis erupted magnetic flux in arcades and dimmings (see Article I; D´emoulin,2008; Mandrini et al. , 2009 for a review).It is clear, however, that the proposed tool for the early preliminary prognosticestimations does not take into account many factors affecting the GMS and FDcharacteristics. Consequently, as in Articles I and II, we do not pursue exactestimates of the parameters of GMSs and FDs and focus instead on their possibleimportance. In practice, our tool should be the starting point of a complex ofcomprehensive forecasting tools, which would consider information on near-the-Sun CMEs, various models of eruptions and drag of ICMEs propagating in the
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 17 hertok, Grechnev, and Abunin solar wind, stereoscopic observations, estimation of a probable sign of the Bz component, and others (see, e.g. , Gopalswamy, Tsurutani, and Yan, 2015, andreferences therein). Acknowledgments
We appreciate the painstaking work of the anonymous reviewer forvaluable remarks and recommendations that significantly helped us to bring this article to itsfinal form. The authors thank the SOHO/EIT and MDI and SDO/AIA and HMI teams fortheir open data used in our study. SOHO is a project of international cooperation betweenESA and NASA. SDO is a mission of the NASA’s Living With a Star (LWS) Program. Weare grateful to A.V. Belov for his assistance and useful discussions. This research was partiallysupported by the Russian Foundation of Basic Research under grant 14-02-00367.
Disclosure of Potential Conflicts of Interest
The authors claim that they have no conflicts of interest.
References
Belov, A.V.: 2009,
Proc. IAU Symp. 257 , 439.
DOI
Belov, A., Baisultanova, L., Eroshenko, E., Mavromichalaki, H., Yanke, V., Pchelkin, V., etal. : 2005,
J. Geophys. Res. , A09S20.
DOI
Bothmer, V., Zhukov, A.: 2007, In: Bothmer, V. and Daglis, I.A. (eds.)
Space Weather - Physicsand Effects , 31.
DOI
Cane, H.V.: 2000,
Space Sci. Rev. , 55. DOI
Chertok, I.M., Grechnev, V.V.: 2005,
Solar Phys. , 95.
DOI
Chertok, I.M., Grechnev, V.V., Belov, A.V., Abunin, A.A.: 2013,
Solar Phys. , 175(Article I).
DOI
Chertok, I.M., Abunina, M.A., Abunin, A.A., Belov, A.V., Grechnev, V.V.: 2015,
Solar Phys. , 627 (Article II).
DOI
Couvidat, S., Schou, J., Hoeksema, J.T., Bogart, R.S., Bush, R.I., Duvall, T.L.: 2016,
SolarPhys. , 1887.
DOI
D´emoulin, P.: 2008,
Ann. Geophys. , 3113. DOI
Delaboudini`ere, J.-P., Artzner, G.E., Brunaud, J., Gabriel, A.H., Hochedez, J.F., Millier, F., et al. : 1995,
Solar Phys. , 291.
DOI
Domingo, V., Fleck, B., Poland, A.I.: 1995,
Solar Phys. , 1.
DOI
Gopalswamy, N., Tsurutani, B., Yan, Y.: 2015,
Progress in Earth and Planetary Science , ,13. DOI
Gopalswamy, N., Yashiro, S., Xie, H., Akiyama, S., M¨akel¨a, P.: 2015,
J. Geophys. Res. ,9221.
DOI
Gosling, J.T.: 1993,
J. Geophys. Res. , 18937.
DOI
Grechnev, V.V., Lesovoi, S.V., Smolkov, G.Y., Krissinel, B.B., Zandanov, V.G., Altyntsev,A.T., Kardapolova, N.N., Sergeev, R.Y., Uralov, A.M., Maksimov, V.P., Lubyshev, B.I.:2003,
Solar Phys. , 239.
DOI
Hanaoka, Y., Shibasaki, K., Nishio, M., Enome, S., Nakajima, H., Takano, T., Torii, C.,Sekiguchi, H., Bushimata, T., Kawashima, S., Shinohara, N., Irimajiri, Y., Koshiishi, H.,Kosugi, T., Shiomi, Y., Sawa, M., Kai, K.: 1994,
Proc. Kofu Symp. , 35.Harra, L.K., Mandrini, C.H., Dasso, S., Gulisano, A.M., Steed, K., Imada, S., et al. : 2011,
Solar Phys. , 213.
DOI
Hudson, H.S., Cliver, E.W.: 2001,
J. Geophys. Res. , 25199.
DOI
Kahler, S.: 1977,
Astrophys. J. , 891.
DOI
Kochanov, A.A., Anfinogentov, S.A., Prosovetsky, D.V., Rudenko, G.V., Grechnev, V.V.: 2013,
Publ. Astron. Soc. Japan . DOI
Krymskii, G.F., Kuz’min, A.I., Krivoshapkin, P.A., Samsonov, I.S., Skripin, G.V., Transkij,I.A., Chirkov, N.P.: 1981,
Kosmicheskie luchi i solnechnyi veter (Cosmic Rays and SolarWind) , Novosibirsk: Nauka.
SOLA: transition_prep.tex; 17 September 2018; 7:49; p. 18 iagnostics of the Geoeffectiveness of Solar EruptionsLemen, J.R., Title, A.M., Akin, D.J., Boerner, P.F., Chou, C., Drake, J.F., et al. : 2012,
SolarPhys. , 17.
DOI
Liu, Y., Hoeksema, J.T., Scherrer, P.H., Schou, J., Couvida, C., Bush, T.L., et al. : 2012,
SolarPhys. , 295.
DOI
Mandrini, C.H., Nakwacki, M.S., Attrill, G., van Driel-Gesztelyi, L., Dasso, S., D´emoulin, P.:2009, In: Gopalswamy, N., Webb, D.F. (eds.),
Universal Heliospheric Processes, IAU Symp. , 265.
DOI
Miklenic, C.H., Veronig, A.M., Vrˇsnak, B.: 2009,
Astron. Astrophys. , 893.
DOI
Pesnell, W. D., Thompson, B.J., Chamberlin, P.C.: 2012,
Solar Phys. , 3.
DOI
Qiu, J., Hu, Q., Howard, T.A., Yurchyshyn, V.B.: 2007,
Astrophys. J. , 758.
DOI
Reinard, A.A., Biesecker, D.A.: 2008,
Astrophys. J. , 576.
DOI
Reinard, A.A., Biesecker, D.A.: 2009,
Astrophys. J. , 914.
DOI
Richardson, I.G., Cane, H.V.: 2011,
Solar Phys. , 609.
DOI
Riley, P., Ben-Nun, M., Linker, J.A., Mikic, Z., Svalgaard, L., Harvey, J., et al.
R.: 2014,
SolarPhys. , 769.
DOI
Scherrer, P.H., Bogart, R.S., Bush, R.I., Hoeksema, J.T., Kosovichev, A.G., Schou, J., et al. :1995,
Solar Phys. , 129.
DOI
Scherrer, P.H., Schou, J., Bush, R.I., Kosovichev, A. G., Bogart, R. S., Hoeksema, J. T, et al. :2012,
Solar Phys. , 207. DOI: 10.1007/s11207-011-9834-2Sterling, A.C., Hudson, H.S., Thompson, B.J., Zarro, D.: 2000,
Astrophys. J. , 628.
DOI
Svalgaard, L., Sun, X.: 2016, http://hmi.stanford.edu/hminuggets/?p=1510 .Thompson, B.J., Plunkett, S.P., Gurman, J.B., Newmark, J.S., St. Cyr, O.C., Michels, D.J.:1998,
Geophys. Res. Lett. , 2465. DOI
Tripathi, D., Bothmer, V., Cremades, H.: 2004,
Astron. Astrophys. , 337.
DOI
Uralov, A.M., Grechnev, V.V., Rudenko, G.V., Myshyakov, I.I., Chertok, I.M., Filippov, B.P.,Slemzin, V.A.: 2014,
Solar Phys. , 3747.
DOI
Watson, F.T., Penn, M.J., Livingston, W.: 2014,
Astrophys. J. , 22.
DOI
Yashiro, S., Gopalswamy, N., M¨akel¨a, P., Akiyama, S.: 2013,
Solar Phys. , 5.