Magnetospheric effects in cosmic rays

during the unique magnetic storm in November 2003

Belov A., Baisultanova L., Eroshenko E., Yanke V. (IZMIRAN, Russia),

Mavromichalaki H. (Athens University, Greece),

Pchelkin V. (PGI, Russia)

Cosmic ray variations of magnetospheric origin during the severe magnetic storm on 20 November 2003 are selected from ground level observation data by means of the global survey method. Planetary distribution of the cut off rigidity variations during this disturbed period was obtained on this basis. For this event a correlation between Dst index and cut off rigidity variation was defined for each cosmic ray station. The most essential shift in cutoff rigidities occurred while Dst index was around -465 nT. Geomagnetic effect in cosmic ray intensity at some stations reached 6-8%. Cutoff rigidity variations were also calculated using the last model of magnetosphere from Tsyganenko 2003. This magnetospheric effect seems to be the greatest one over the history of neutron monitor observations. Maximum changes of geomagnetic cutoff rigidities were recorded this time at unusual low latitudes corresponded to about 7-8 GV cutoff rigidity. The results can be used in the neutron monitor data correction for magnetospheric effects and for the more accurate modeling as well of the magnetospheric current system during strong geomagnetic storms.

Introduction.

Disturbances of the Earth’s magnetic field during the magnetic storms can cause essential changes of the charged particle trajectories in the magnetosphere, up to allowed trajectories become forbidden, and conversely. It makes two main consequences for ground level observations: 1) the effective cut off tresholds are changing; 2) the effective asymptotic patrticle directions and, hence the reception coefficients for different stations are also changing. Both these consequences are important for solar cosmic rays (CR), whereas for galactic CR the first effect is dominated. Magnetosphere effect associated with the cut off rigidity changes may be sufficiently large to distort essentially cosmic ray variations on the fixed station observed or even to change completely its behavior. As an example, a magnetosphere effect during magnetic storm on 20 November 2003 may be considered (see Fig. 1).

There are some reasons of a special interest to magnetosphere variations. In the first turn these effects are interesting on the physical point of view: creation, evolving and decay of the magnetosphere current systems, global interaction of cosmic radiation with the geomagnetic field. And, secondary, magnetosphere effects are important from methodological side since they hinder to study primary CR variations and should be excluded from the initial data. Large magnetosphere effects are usually observed simultaneously with the big modulation effects in cosmic rays since they both have the common reasons.

Cosmic ray variations caused by the cut off rigidity changes during the big magnetic storm, have been already studied in many papers [1-9]. Nevertheless, a set of important tasks still exists in this area. Some of them are the following:

1) To study all large (>100) magnetic storms, to develop a method of correction for geomagnetic effects, and to clear CR data of the world wide neutron monitor network from magnetosphere variations. We expect finding a quantitative relation between and possible for each station after study of sufficient number of magnetic storms. These results can be used for magnetosphere effect correction data from NMs if only as the first approach.

2) Checking of the current system models for different steps of the magnetic storm evolving. Incorporation directly cosmic ray data for this analysis is important to study the global effect of the current systems on the particle trajectories in such magnetic fields. Moreover, these studies are useful to be carried as for initial phase of magnetic storm associated with currents in the magnetopause, when cut off rigidity increases relatively to the quite level, so in the main phase, when cut off rigidity decreases significantly (Baisultanova et al., 1995).

In this paper we intended to study in details magnetosphere effects on the example of severe magnetic storm on 20 November 2003.

Events on the Sun and in the interplanetary space in November 2003

Two active regions at the Sun were the most effective on 18 November 2003: 501 (484) and 508 (486). The last gig flare in group 508, accompanied by the powerful coronal mass ejection (CME), was observed on 18 November at the eastern limb (M4., onset at 09:23 UT, maximum at 10:11 UT). At the same time in the group 501 two long lasted flares occurred in the center of disk (M3.2/2N N00W18, onset at 07:16 UT, maximum at 07:54 UT; M3.9/, onset at 08:12 UT, maximum at 08:31 UT), which were also followed by powerful and extremely effective CMEs. Severe magnetic storm associated with the flares on 18 November (as minimum with two central and possibly with the all three) started on 20 November. After shock arrival (at 07:28, SOHO; SSC 08:04) and Earth enter into the long magnetic cloud, the IMF intensity reached 60 nT, and almost this value had its negative Bz-component. So, geomagnetic activity at the end of 20 November increased up to the level of severe magnetic storm, -index fall down to -465 nT – it was lower only one time on 13-14 March 1989. Aurora was observed even in the southern Europe (Athens).

Data and Method.

Analysis was performed with the hourly data from neutron monitors of the worldwide network. Data from 39 NMs have been used: 15 high latitudinal (<1.2 GV), 22 – mid latitude and 2 sub equatorial (>10 GV) stations. Full listing of the used stations is presented in the Acknowledgements. Operatively calculated Dst indexes for November 2003 were taken from the WDC-C2 server: (http://swdcwww.kugi.kyoto-u.ac.jp/dstdir/).

Analysis and calculations were based on the global survey method (Krymsky et al., 1966), which is conceptually a complicated version of the method of spherical analysis. In fact it combines three methods. Method of coupling coefficients allows the observed ground level CR variations to be connected with expected variations on the boundary of atmosphere. Method of trajectory calculations in the magnetosphere allows these variations to be connected with expected variations beyond of magnetosphere. And, on the final step, method of spherical analysis is used to select spherical harmonics important for a specific task and further analysis. Different versions of this method have been evolved and improved on different steps of data processing. We used as a basis the version of this method described in (Baisultanova et al., 1987; 1995). In the main equations the following notations have been used:

- a spectrum of CR variations, where is the amplitude of the zero harmonic of CR variations, - spectral index;

- three components of the first harmonic of CR variation in decart coordinate system; - are reception coefficients for each component respectively; - is coupling function for detector , localized at the level in the point with geomagnetic cut off rigidity . Observed CR variations in common can be written as following:

+++, (1)

where is magnetosphere variation, and mean isotropic and anisotropic CR variations out of magnetosphere, and is residual fluctuation which describes possible apparatus variations and non adequate used model. In the assumption data are free from meteorological effect and using only the first spherical harmonic, the system of spectrographic equations can be written for CR variations at point:

+++, (2)

The system is solving by the least square method of minimization of sum of the squares of residual variations relatively to the unknown parameters , and components of vector of anisotropy. Two approaches are possible for magnetosphere variations. In one case the model dependence from rigidityis specified, for example, as it was done in (Dvornikov et al., 1988): . In this case the system is solved for the set of parameters , and ,, and . It is clearly, the assign of latitudinal dependence on limits in advance the possibilities of this method. Cut off rigidity variations caused by the magnetosphere current ring during the main phase of magnetosphere storm have a minor longitudinal dependence because of the ring symmetry. On the contrary, during the initial phase of the magnetic storm they have a significant longitudinal dependence, since current distribution on the daily side of magnetosphere differs considerable from the night distribution.

To find latitudinal dependence of the cut off rigidity variations, another approach, not based on the model of rigidity dependent, has been used. The sense of this approach is that a system (2) has been solved without term of magnetosphere variations; these variations were included into residual errors. In this case we can write:

, (3)

where - is a contribution at a sacrifice of non adequate model (spectrum, harmonics of higher order), - high frequency, statistical component, - low frequency component (drift of apparatus). We can neglect contribution from the last two terms applying corresponding filters. If three last adds are negligible in compare with magnetosphere variations, then, , i. e. all residual error may be attributed to magnetosphere effect. In this case we can write:

(4)

Under this approach the cut off rigidity variation at each station are defined independently of one another. To avoid the interplay between harmonics a special care was taken under system (3) solving.

Results and discussion. It can be seen from the first view on the CR data from different neutron monitors (Fig. 1) that Forbush decrease was moderate and with sufficiently hard spectrum despite of extremely severe magnetic storm in this period. Magnetosphere effect in CR was maximal on the low latitude stations, but not on the mid latitude, as it is often observed. It was significant by the amplitude, so, Forbush decrease on these stations was masked completely by the

magnetosphere effect, as it is indicated in Fig. 1, where uncorrected and corrected for the magnetosphere effect CR variations are presented for stations Athens and Potchefstroom and are compared with the same variations on high latitude stations Apatity and McMurdo. Cut off rigidity variations were calculated for each station for different times during the storm by the method above mentioned. This result is plotted in Fig. 2 for Athens and Jungfraujoch stations, and for all other stations it is presented in Appendix A1. In this Figure a comparison of obtained with -index is also presented. It is not an approximation, but visual scale concurrence. One can see very high correlation over the whole considered period. Despite of usually Jungfraujoch station in 2 times is more sensitive to geomagnetic effects than Athens, in this event Athens reveals geomagnetic effect in twice bigger. As it will be shown below, such an effect is caused by the peculiarity of the storm on 20 November 2003, namely, by specific space distribution of the current system. In Fig. 3 the same results are presented as correlation dependences (for the all stations these dependences are collected in the Appendix A2 and A3). Two regions are clearly pronounced in this Figure: of small and large (>50 nT) Dst index. Within the first region an accuracy of can be easy estimated (as 0.1 GV for each station). Within the region of large Dst index with Dst amplitude >50 nT approximately linear dependence on Dst is observed. For Athens station a regression coefficient is equal to 0.00267, for Jungfraujoch it is 0.00178. Since more accurate calculations assume to be carried out after the complete data set collection, we did not perform a detailed analysis, including the estimation of precise of obtained approximations. Note that the empiric dependence of Dst index with Dst amplitude obtained for the Erevan station (Martrosyan G., 2004) corresponds rather well to our results.

Latitudinal dependences of the cut off rigidity variations were found for each hour starting from the moment of shock arrival and up to final recovery of the magnetosphere. These results are presented in the Appendixes A4 and A5, and dependence for maximum evolving of the magnetic storm is plotted in Fig. 4.

Figure 4 shows a comparison between experimental and calculated values of Rc changes for moment of the maximum effect at 19.30 UT. Calculations were performed by the last “storm” model T01s of the magnitosphere [12]. The method is described in [13]. The trajectories were calculated from the main cone to the Stormer cone adding all allowed intervals (i.e. for the plane spectrum of CR). The step of calculations was 0.002 GV. A duration of trajectory calculations for quasi-trapped particles was chosen to reach vicinity of the asymptotic value. The model was tested for the rather quite period at 6:30 UT on November 20. For this point the classical package T89 and the new T01s give very close values. Changes of cut off rigidities were determined relatively this time moment. Because experimental points have been determined for the main magnetic field model IGRF-1990 they shifted a little by Rc. Clear that for rigidities >6 GV there is a good agreement between experimental and calculated values, moreover, without any normalization. Possibly, the model is not adequate for the greatest magnetospheric disturbances; this causes a discrepancy for lower rigidities. In favor of our method note that performing the same analysis for other magnetic storms [7] we have got the classical latitudinal dependence of Rc changes with maximum at 3-4 GV. Latitudeal dependence of cut off rigidity before, in the maximum of the magnetic storm evolving and after storm you can see in Appendix A6.

In [14] was examined the azimuthal electric currents flowing in the magnetospheric. The currents were extracted from the magnetic databases of Fairfield et al. [1994] as well as Tsyganenko et al. [2003] statistically, the measurements being accumulated in spatial bins. They have also analyzed several models of the magnetic field in the magnetosphere [Tsyganenko et al., 2003; Alexeev et al., 2003; Maltsev and Ostapenko, 2001]. The model currents are compared with those obtained from the databases. Azimuthal currents in the magnetospheric layer -3 < z_SM < 3 R_E obtained from observations. They extracted the currents from the magnetic databases statistically, by accumulating the measurements in spatial bins. Clear, that the maximum amplitude value of the Dst module obtained from different models is 140 nT and calculations of for giant magnetic storms with Dst values of several hundreds nT by these models would be not correct.

As it was mentioned, a specific feature of this event is that maximal magnetosphere effects was recorded at low latitude stations, instead of mid latitude as usual. In this case maximum in latitudinal distribution of the cut off rigidity variations is shifted significantly to the bugger rigidity, and is found approximately around 8-9 GV (instead of usual 4-5 GV). It means that ring current which accordingly to the simplest model (Treiman, 1953) is distributed by latitude proportionally to cosines of this latitude and carries in western direction, is maximally closed to Earth in this case and located at 3R from the Earth center. In the magnetic storms when maximum in latitudinal distribution of the cut off rigidity variations is nearly 4-5 GV, current system is placed at about 5 Earth radii.

The errors are given in this Fig.4 as those derived from the system equation solving. In fact the errors may be apparently caused by some other sources which are more difficult estimated. In particular, we don’t know the exact coupling functions around the geomagnetic cut off rigidity for each station. In Fig. 6 coupling functions for several stations used in our calculations, are presented. Penumbra region, inclined incident particles as well lead to the blur of coupling function nearly to , hence some effective values necessary to be used to account properly this blur. This is the biggest uncertainty in our task. The observed dispersion of in Fig. 4 seems to be related directly to this uncertainty. Besides, in some times an additional error is possible because of unaccounted 2-nd harmonic.Magnetosphere variation in CR is defined as the productive, so the value of coupling function under cut off rigidity causes a sensitivity station to the magnetosphere effect. In Table A1 the values of in usual case in units are given for different stations. One can see from this Table A1 that in usual case Jungfraujoch station is about in twice more sensitive to magnetosphere effect than Athens, whereas in the event on 20 November 2003 Athens showed magnetosphere effect in two times bigger than Jungfraujoch. This is related with particular latitudinal distribution of the cut off rigidity variations in this event.

Main conclusions.

1) At the beginning of extreme magnetic storm on 20 November 2003 a small magnetosphere effect in cosmic rays was recorded, whereas an exclusively large effect was derived during the main phase of this storm. This allowed the latitudinal distribution of the cut off rigidity variations to be obtained for each hour during the main and recovery phases of this magnetosphere storm very useful for analysis a dynamic of evolving and damping out of the ring current systems.

2) The ring current system during magnetic storm on 20 November 2003 was at more close distance from Earth (apparently about 3 Earth’s radii) than it usually observed. As a consequence, maximal magnetosphere effect in CR was recorded at lower latitudes, but not at the mid latitude stations as usual. Maximum changes of the geomagnetic cut off rigidity were shifted due to this anomalous from 3-4 GV (usually) to 7-8 GV.

3) The calculations of Rc changes performed for the last "storm" model T01s of the magnetospheric magnetic field [12] shows a good agreement between experimental and theoretical values for rigidities > 6 GV. Possibly, the model is not adequate for the greatest magnetospheric disturbances, this causes a discrepancy for lower rigidities.

4) Further developing and extended variant of this work, whether another events are analyzed, can be found on http://cr0.izmiran.rssi.ru/GeoMagCR/main.htm.

Acknowledgements

This work n operating of the Russian CR stations is partly supported by Russion FFR Grants 02-02-16992, 03-07-90389 and 04-02-16763; for USA stations - NSF Grant ATM-0000315. The authors thank the collaborators from all CR stations which data were taken for analysis: Alma-Ata, Apatity, Barensburg, Calgary, Cape Shmidt, Climax, Erevan, Fort Smith, Haleakala, Hermanus, Inuvik, Irkutsk, Jungfraujoch, Kiel, Larc, Lomnicky Stit, McMurdo, Magadan, Moscow, Nain, Norilsk, Novosibirsk, Newark, Oulu, Potchefstrom, Peawanuck, Tixie Вay, Rome, Sanae, South Pole, Thule, Tsumeb, Yakutsk, ESOI, Mexico, Kergelen, Terra Adelia and Beijing, Tibet, Mawson, Kingston.

Reference

1. Debrunner H., Flueckiger E., Mandach H., Arens M. // Planetary and Space Science. 1979. V. 27. P. 577.

2. Flueckiger E. O., Smart D. F., Shea M.A. // Proc. 17th ICRC. Paris. 1981. V.4. p. 244.

3. Fluckiger E., Smart D., Shea M. // Proc. 20-th ICRC. Moscow. 1987. V. 4. 216.

4. Dvornicov V., Sdobnov V., Sergeev A. // Proc. 20-th ICRC. Moscow. 1987. v. 4, 220.

5. Baisultanova L., Belov A., Dorman L., Yanke V. // Proc. 20-th ICRC. 1987. V. 4. 231.

6. Dvornikov V., Sdobnov V., “Modification of the method for spectrographic global surveyfor studing variation I theplanetary system of geomagnetic cutoff rigidities”, Izv. AN SSSR, Ser. Phys. Vol. 55(10), p. 1991, 1988.

7. Baisultanova L., Belov A., Yanke V. // 1995. Proc 24-th ICRC. Roma. v. 4. 1090.

8. Sdobnov V., Dvornicov V. Lukovnikova .A.,Osipova N.A. // Sun-Earth physics, Irkutsk, 2002. вып. 2. p. 230.

9. Dvornikov V., Sdobnov V. // Intern. JGA. 2002. V. 3. No. 3. Р. 1-11. February 2002.

10. Krymskiy G.F. at al., “Cosmic ray distribution and reception vectors of detectors, 1”, G&A, 6, 991, 1966.

11. Treiman S.B. // Phys. Rev.. 1953. V. 89(1). Р. 130.

12. Tsyganenko N.A., Singer H.J., Kasper J.C. // J. Geophys. Res.. 2003. V. 108(A5). 1209. doi:10.1029/2002JA009808.

13. Pchelkin V.V., Vashenjuk E.V. // Izv. AN SSSR, Ser. Phys. Vol. 65(3), p. 416, 2001.

14. Maltsev Y. P., Ostapenko A. A., Pchelkin V. V. // “Predictions of the magnetospheric electric currents during superstorms”, Izv. AN SSSR, Ser. Phys. In press, 2004.

15. Fairfield, D. H., N. A. Tsyganenko, A. V. Usmanov, and M. V. Malkov, A large magnetosphere magnetic field database, J. Geophys. Res., 99, No A6, 11,319-11,326, 1994.

16. Tsyganenko, N. A., H. J. Singer, and J. C. Kasper, Storm-time distortion of the inner magnetosphere: How severe can it get? J. Geophys. Res., 108, No A5, 10.1029/2002JA009808, 2003.

17. Alexeev, I. I., E. S. Belenkaya, S. Yu. Bobrovnikov, and V. V. Kalegaev, Modelling of the electromagnetic field in the interplanetary space and in the Earth's magnetosphere, Space Science Reviews, 107, Issue 1-2, 7-26, 2003.

18. Maltsev, Yu. P., and A. A. Ostapenko, Model of the magnetospheric magnetic field (in Russian), Geomagnetism and Aeronomy, 41, No 6, 761-765, 2001.;Maltsev, Y. P. and A. A. Ostapenko, Azimuthally asymmetric ring current as a function of Dst and solar wind conditions, accepted by Ann. Geophys., 2004.

19. Martrosyan G., // “Magnetospheric effect of cosmic rays and rigidity variations”, Izv. AN SSSR, Ser. Phys. In press, 2004.

Appendix A1

Appendix A6. Latitudinal dependence of cut off rigidity before, in the maximum of the magnetic storm evolving and after storm at 20 November 2003.

Table 1. Listing of the most sensitive stations to the geomagnetic effects.

	Lat	Long	Alti- tude,m	H0, mb	Rc, GV	W(Rc), %/GV
Jungfraujoch	46.55	7.98	3550	643	4.48	10.62
Irkutsk3	52.28	104.02	3000	715	3.66	9.49
Climax	39.37	-106.18	3400	685	3.03	9.36
Alma-B	43.14	76.60	3340	675	6.69	9.10
Erevan3	40.50	44.17	3200	700	7.60	8.33
Irkutsk2	52.28	104.02	2000	800	3.66	8.29
Erevan	40.50	44.17	2000	800	7.60	7.36
Potchefstrom	-26.68	27.92	1351	869	7.30	6.82
Mexico city	19.33	-99.18	2274	794	9.53	6.59
ESOI	33.30	35.78	2025	800	10.00	6.37
Alma-A	43.25	76.92	806	938	6.66	6.36
Irkutsk	52.10	104.00	433	965	3.66	6.18
Tsumeb	-19.20	17.60	1240	880	9.29	6.00
Hermanus	-34.42	19.22	26	1013	4.90	5.89
Huancayo	-12.03	75.33	3400	704	13.45	5.79
Rome	41.90	12.50	60	1009	6.32	5.75
Haleakala	20.72	-156.27	3052	724	12.91	5.72
Athens	37.97	3.72	40	980	8.53	5.22
Tibet	30.11	90.53	4300	606	14.10
Beijing	40.04	116.19	48	1000	9.56