Attenuation of GPS scintillation in Brazil due to magnetic storms

SPACE WEATHER, VOL. 6,, doi:10.1029/2006sw000285, 2008 Attenuation of GPS scintillation in Brazil due to magnetic storms E. Bonelli 1 Received 21 September 2006; revised 15 June 2008; accepted 16 June 2008; published 4 September 2008. [1] Amplitude scintillations in satellite signals can cause errors in communications, because of signal fading, but can be very useful for scientists trying to improve their understanding of the physics of the ionosphere. Usually, magnetic storms are expected to affect the ionosphere in such way as to increase ionospheric irregularities responsible for scintillations. To help change the view of scientists and engineers, in this respect, we show that amplitude scintillation on GPS signals show dramatic decrease during selected magnetic storms, at Brazilian GPS stations. These stations are located on magnetic latitudes that go from equatorial (São Luís) to low-latitude (São José dos Campos and Cachoeira Paulista) so that a region of several thousand kilometers is represented by the data. We present 4 months of data chosen from 2003 to 2005 to represent the strongest storms during each scintillation season. Although there is lack of data for some days from the different stations, it is possible to see, especially for the Halloween Storm (October 2003), that scintillations are attenuated in this wide range of latitudes. During magnetically calm periods scintillations are strong, in this region, from August to March, during solar maxima. Although the data are clear about the attenuation of scintillations during greater magnetic storms, it is not possible to easily conclude which physical mechanism was responsible for this phenomenon, even with the aid of more detailed data like Dst and AE. Citation: Bonelli, E. (2008), Attenuation of GPS scintillation in Brazil due to magnetic storms, Space Weather, 6,, doi:10.1029/2006sw000285. 1. Introduction [2] The Natal GPS station is located at 5.84 S and 35.20 W, geographic coordinates. The magnetic declination by the end of 2003 was 22 and the magnetic dip was 21.6, which corresponds to magnetic latitude (dip latitude) of 10. For Natal, the dip angle has being changing by about 1 every 3 years, since 1945, when it was located at the magnetic equator. Other Brazilian stations considered in this paper are, from low-magnetic to midmagnetic latitudes: Manaus (+5 ), São Luís ( 1.5 ), and São José dos Campos ( 17 ). The numbers in the parenthesis indicate their approximate dip latitudes, or magnetic latitudes. The local time for Manaus is UT minus 4 h, while for the other stations it is UT minus 3 h. The GPS scintillations were obtained using the L1 frequency (1.575 GHz). It is known that, at this frequency, the signal is most sensitive to irregularities, at zenith, of scale size around 400 m [Yeh and Liu, 1982; Kintner et al., 2001]. The scintillation index, S 4, used to measure the effects of ionospheric irregularities on the signal is defined as the dispersion of power divided by 1 Departmento de Física, Universidade Federal do Rio Grande do Norte, Natal, Rio Grande do Norte, Brazil. average power [Briggs and Parkin, 1963]. The method used in analyzing the data is to calculate an average S 4 index for all available satellites, for the night period, and compare it with the sum of 3-h K p indices for several hours before sunset. The data include four major storms for which we had simultaneous scintillation data. In Natal the scintillation season goes from August to April, because of both the inclination of the axis of the earth with respect to the ecliptic and to the declination of the geomagnetic field. During this period of months, the sunset at both hemispheres occur at the extremities of the same field lines, so dynamo electric fields across magnetic field lines in one hemisphere cannot be shorted out in the other hemisphere, as it would be if one end of the field lines were in the dark and the other still in sunshine, immersed in high conducting plasma. 2. Data Analysis [3] The scintillation data is obtained at a rate of 50 samples per second for each available satellite. Further data processing produces the scintillation index, S 4, per Copyright 2008 by the American Geophysical Union 1of9

Figure 1. Scintillations for 28-- 29 October 2003 for satellites passing near the zenith at Natal. For each section the same satellite is represented, for the consecutive nights, with 5 min delay on 29 October. The magnetically calm day is represented by the black line and the disturbed day by the gray line. Note that there are no data gaps for either day, for any of the satellites and day. minute of data, along with 1 min averages of wide band power and narrow band power, for each satellite. [4] To summarize a whole night of scintillations we averaged the 1 min S 4 for all satellites present at that night, from 2100 (near local sunset) to 0600 UT. This average will be referred to as hs 4 i. The number of points used in the computation of hs 4 i shows whether all satellites are well represented in the average. This number lies, in general, between 1000 and 2000 points. [5] To represent the magnetic activity, we use the sum of K p from 0000 to 1800 LT, which will be denoted by P K p. The reason for choosing this time interval for K p is that we plan to forecast GPS scintillations, so that at sunset we could be able to know how much scintillations should be expected at night. For the same reason, we used real-time estimates of K p, from the magnetic data from stations around the world (http://www.sec.noaa.gov/rt_plots/satenv.html). [6] To try to have a more detailed understanding, of what was observed in the above gross average of K p we add also plots of estimated D st and measured AE indexes. The AE indexes are those from World Data Center-Kyoto (http://swdcwww.kugi.kyoto-u.ac.jp/index.html). The estimated D st index (http://lasp.colorado.edu/space_weather/ dsttemerin/dsttemerin.html) is a real-time index, toward which the Kyoto values converge when they become definitive, after some years. 3. Results and Discussion [7] During strong storms, amplitude scintillation might become so intense that the GPS receiver could lose the lock to one or more satellites. This effect is very rare in Natal. Sometimes there is loss of lock onto satellites, but this happens more often during daytime. To show that track is kept of all satellites, during strong scintillations we show (Figure 1), the strong scintillations of 28 October 2003 (in black), superposed to the weak scintillations of 29 October 2003 (in gray) for Natal. The satellites chosen are those that passed near the zenith. From one night to the next, the time a satellite moves into view is about 5 min latter than for the previous day. The path in the sky is the same for both days. In Figure 1, it is clear that the satellites were kept track of during the strong scintillations. Actually, the main result, for this paper, is the gray plot, where scintillations are very weak and a satellite could not be lost because of it. [8] To see the behavior of GPS scintillations during magnetic storms we plot hs 4 i and P K p versus day for each month of data available. We chose months with strong storms and with less days with missing scintillation data. Since the beginning of the GPS era, there were few strong storms and we missed some, so that we present below only four major storms. In what follows, the word storm corresponds to a day or to consecutive days with high P K p, even if there were more than one superposed storms, in terms of D st. Since we are summing up several 3-h K P indices, it is convenient to define what we understand for a high K p sum: a high P K p is considered to be above 24, meaning that the mean 3-h index is about 4, since K p greater than 3+ is supposed to represent disturbed 2of9

Figure 2. Average S 4 for October 2003 for Natal and Manaus and P (K p ) versus day. conditions. In the text below, whenever we mention latitude we refer to the magnetic latitude. A word about São José dos Campos and Cachoeira Paulista. These two stations are very close so that they have approximately the same magnetic dip and declination. This allows us to use data from Cachoeira Paulista, when those of São José dos Campos are missing for a given day. When data from both stations are present, São José dos Campos data are plotted. 3.1. The 2003 Halloween Storm [9] The strongest storm from 2003 to 2005, the Halloween storm, occurred in the period of 29--31 October 2003 (Figures 2-- 4). In Figures 2-- 4, we see that during the 4 days preceding the the onset of the storm, P K p was low, as compared to the storm days. [10] On the storm days, 29-- 31 October, there was a dramatic decrease of hs 4 i for Manaus (Figure 2), São Luís (Figure 3), and Natal. On the other hand, although in São José dos Campos (SJC) and Cachoeira Paulista (Figure 4) there was a decrease in scintillations, this decrease was not as abrupt as in the low-latitude stations. In Figure 1, we show details of the wide band power for satellites that passed near the zenith in the first night during the storm (gray line) and in the previous night (black line). It is very clear that scintillations almost disappeared during the storm, for Natal. [11] The strong scintillation, for 28 October can be attributed to the usual quiet day behavior, at these stations. On the 29 October, both Dst (Figure 5) and AE (Figure 6) show the beginning of a storm, and scintillation diminishes. In this case the absence of scintillation could mean either a washout of the plasma, due to previous strong eastward electric field, due to instant penetration from the auroral zone or a stabilization of the ionosphere due to a strong disturbed dynamo, since the auroral storm occurred several hours before. [12] After the storm, hs 4 i remains low at the beginning of November (see also Figure 7), although P K p is Figure 3. Average S 4 for October 2003 for Natal and São Luís and P K p versus day. 3of9

Figure 4. Average S 4 for October 2003 for Natal and São José dos Campos and Cachoeira and P K p versus day. moderate. Would this mean that the effect of the disturbed period may remain for some days? In this case, the responsible mechanism could be the disturbance dynamo. Actually, a strong auroral storm occurred in the evening of 30 October, as shown in the AE plot (Figure 6), and it is possible that it would have caused the attenuation of scintillation on 1 November. [13] For this period, there is data from other stations [Basu et al., 2005] for scintillations of a geostationary satellite, at 250 MHz. We shall come back to this in the discussion section. 3.2. November 2003 [14] For this month, only 1 day (20 November) stands out from the others, in P K p, although not by much, and we considered it as a magnetically disturbed day (Figure 7). Even for this situation of a short-duration storm, the decrease in hs 4 i is very clear both for Natal, São José dos Campos, and Manaus. Although data for São Luís were not available for this day, we see that the behavior was the same at all latitudes. São José already had weak scintillations when the storm occurred, however, it dropped even further and did not recover before 4 days, unlike Natal and Manaus, which recovered in the next day. The Manaus data was plotted only around 23 November, which is the important day for our reasoning. [15] Analyzing the more detailed behavior of Dst (Figure 8) and AE (Figure 9), for days 19--22, we see that there was a broad maximum of auroral activity during 20 November, while the main phase of the storm ended in the afternoon (Natal local time is UT minus 3 h) of 21 November. In this case it is more reasonable to assume that the inhibition of scintillation was due to a disturbance dynamo westward electric field. 3.3. October 2004 [16] For this month, the maximum of P K p occurred on 13 October with a second maximum on 14 October (Figure 10). This was the mildest storm of all data presented ( P K p 30) and, despite that, there is a pronounced valley in the scintillations of São Luís and Natal, for these days. São José dos Campos scintillations were not affected by this storm. There is no data for Manaus for this period. [17] The Dst data (Figure 11) present two superposed storms. The first starts on the 12 October, by midnight, its main phase ending around noon on the 13 October. In this case, at local sunset, on the 13 October, where the first inhibition of scintillation occurred, the storm, as viewed through Dst, was already in the recovery phase. Usually, in Figure 5. Estimated Dst versus time for October 2003. 4of9

Figure 6. AE index versus time for October 2003. this case, scintillation increases, but this did not happen. To understand what caused the attenuation of scintillation, let us turn to the behavior of the auroral region, represented by the AE index (Figure 12). AE peaked around noon, on the 13 October, and the disturbance dynamo could be the responsible for the diminution of scintillation. This last explanation also applies to 14 October, since AE has a second peak around noon, again, on this day. 3.4. January 2005 [18] Here we have a relatively strong storm which occurred from days 17 to 19, and a minor one from 21 to 22 (Figure 13). The reduction of average S 4 are apparent for Natal and São Luís. At this time of the year, scintillations were weak in São José dos Campos, especially at this time, near the coming solar minimum. Note the increase in scintillations, for 1 day only, in all three stations, when the average K p fell between this large storm and a smaller one. Natal and São Luís seemed to respond to this second storm, too. [19] Regarding the analysis of the behavior of the indexes Dst (Figure 14) and AE (Figure 15) we see that while Dst did not reach very negative values, the auroral zone was very stormy, with many peaks of activity. Attenuation of scintillations could be due to disturbed dynamo, due to any peak in activity, during the day. With respect to 16 January, there was some scintillation since both indexes indicated a quiet time and then, the normal prereversal eastward electric field was the main driver of upward plasma drifts. 4. Discussion and Conclusions [20] Although there is indication [Basu et al., 2001] that magnetic storms can either enhance or inhibit the formation of ionospheric irregularities responsible for scintillations, the results shown here indicate that for the four selected major storms the inhibition always occur be it because of prompt penetration of westward electric field, due to disturbance dynamo, or due to a wash out of the plasma. [21] Detailed analysis, as done in the reference above, and references therein, are very important to show how prompt penetration electric fields [Scherliess and Fejer, 1997] and disturbance dynamo fields [Fejer and Scherliess, 1997] might control the formation of the irregularities. In this work, however, all conclusions seemed to lead to the disturbance dynamo interpretation of the scintillations. One must be careful, since the auroral behavior, shown by AE, seems to explain, also, the attenuation of scintillations during the Halloween storm, when the ionosphere Figure P 7. Average S 4 for November 2003 for Natal and São José dos Campos and Manaus and Kp versus day. 5of9

Figure 8. Estimated Dst versus time for November 2003. could have been washed out because of a strong eastward electric field. [22] Using 250 MHz scintillation data from a geostationary satellite, whose name is not mentioned by the authors, from 4 stations, [Basu et al., 2005] suggested that there is a connection between the time of the penetration of electric fields from high latitudes and the time of local sunset. For this reason, 3 stations had scintillations because they fulfilled the sunset requirement. Actually, 2 stations had this behavior, since one of them was near or inside the storm expanding aurora oval. The Brazilian stations considered here, however, had the same storm behavior as the fourth station, Ancon, i.e., no scintillation. For Natal, the sunset time suggests it should be affected, especially on 31 October, when the penetration occurred during the time of local sunset. The other stations of this paper, Manaus, São Luís e SJC are reasonably close to Natal, in latitude, to be considered equivalent, in terms of sunset time. In fact, they behave in a very similar fashion regarding the behavior of scintillations during the magnetic storms that were considered here. With regards to VHF scintillations, it is known, since 1983, that they are also attenuated during magnetic storms in Natal [Medeiros et al., 1983]. [23] Using L-band scintillations from a geostationary satellite [Groves et al., 1997], showed that at Ascension Island (dip latitude 15 ) scintillations suffered more attenuation as the K p index got higher. Considering these two cases of attenuation, in L-Band and VHF, which occurred a long time ago, in terms of the variation of dip angle and declination in the region, it is conforting to know that the the same physics seems to be in action, meaning that recent results are similar to the ones of some years ago. To detect a variation of the dependence of the phenomenon with the geomagnetic field variation in the region, one would have to go further back in time. That would be one of our next tasks since, in Natal, VHF scintillation data is available since 1976. [24] The 4 months of scintillation data presented do not represent the whole period of important storms from October 2003 to January 2005. In order to show that, despite this, the conclusions are of significance, let us discuss the strong storms for which there was no scintillation data at Natal, this station being the guide on what dates to select from this and from the other stations. In July 2004, there was a strong storm. This time of the year, however, is outside the scintillation season for the Brazilian stations, which starts in August and ends in April, as explained in the introduction. The other missing period is the one corresponding to storm(s) that occurred from days 8--10 November in 2004. This was an important period at the center of the scintillation season, and it is unfortunate that we have no scintillation data, in Natal, for that period. On the overall, there were six important periods of storms, during the data spam, of which we have presented data for four of them, but the July 2004 storm should not count, since there would be no scintillations to register. For this reason, we have scintillation data for most of the important storms from 2003 to 2005, which means that the results above should repeat for the coming storms, outside solar minima. A continuation of this work will be the analysis of the solar minimum behavior of scintillations during magnetic storms. In this case, only Natal, São Luís and Manaus should exhibit scintillations. [25] Works about GPS scintillations in São Luís and Cachoeira Paulista make only brief comments about storm effects [e.g, Kil et al., 2000; Rodrigues et al., 2004]. Figure 9. AE index versus time for November 2003. 6of9

Figure 10. Average S 4 for October 2004 for Natal and São José dos Campos and São Luís and P K p versus day. Figure 11. Estimated Dst versus time for October 2004. Figure 12. AE index versus time for October 2004. 7of9

Figure 13. Average S 4 for January 2005 for Natal and São Luís and São José dos Campos and P K p versus day. Figure 14. Estimated Dst versus time for January 2005. Figure 15. AE index versus time for January 2005. 8of9

[26] Acknowledgments. The GPS card and antennae were kindly donated by Cornell University, through Dr. Paul Kintner of the Department of Electrical Engineering in 1996. He is also responsible for the Brazilian net of observatories, coordinated nationwide by Dr. Eurico Rodrigues de Paula to whom I am greatly indebted for providing the data for stations other than Natal. I also thank Dr. Theodore Lyman Beach for the software to collect and treat the data. This work has been supported by continuous grants from the FAPERN/ Rio Grande do Norte and CNPq Brazilian agencies. References Basu, S., et al. (2001), Ionospheric effects of major magnetic storms during the International Space Weather Period of September and October 1999: GPS observations, VHF/UHF scintillations, and in situ density structures at mid and equatorial latitudes, J. Geophys. Res., 106, 30,389-- 30,413. Basu, S., S. Basu, K. M. Groves, E. MacKenzie, M. J. Keskinen, and F. Rich (2005), Near-simultaneous plasma structuring in the midlatitude and equatorial ionosphere during magnetic superstorms, Geophys. Res. Lett., 32, L12S05, doi:10.1029/2004gl021678. Briggs, T. L., and I. A. Parkin (1963), On the variation of radio star and satellite scintillations with zenith angle, J. Atmos. Terr. Phys., 25, 339--365. Fejer, B. G., and L. Scherliess (1997), Empirical models of storm time equatorial zonal electric fields, J. Geophys. Res., 102, 24,047-- 24,056. Groves, K. M., et al. (1997), Equatorial scintillation and systems support, Radio Sci., 32(5), 2047-- 2064. Kil, H., P. M. Kintner, E. R. de Paula, and I. J. Kantor (2000), Global Positioning System measurements of the ionospheric zonal apparent velocity at Cachoeira Paulista in Brazil, J. Geophys. Res., 105(A3), 5317-- 5327. Kintner, P. M., H. Kil, T. L. Beach, and E. R. Paula (2001), Fading timescales associated with GPS signals and potential consequences, Radio Sci., 36(4), 731-- 743. Medeiros, R. T. D., M. A. Abdu, and I. J. Kantor (1983), A comparative study of VHF scintillation and spread F events over Natal and Fortaleza in Brazil, J. Geophys. Res., 88(A8), 6253-- 6258. Rodrigues, F. S., E. R. de Paula, M. A. Abdu, A. C. Jardim, K. N. Iyer, P. M. Kintner, and D. L. Hysell (2004), Equatorial spread F irregularity characteristics over São Luís, Brazil, using VHF radar and GPS scintillation techniques, Radio Sci., 39, RS1S31, doi:10.1029/ 2002RS002826. Scherliess,L.,andB.G.Fejer(1997),Stormtimedependenceof equatorial disturbance dynamo zonal electric fields, J. Geophys. Res., 102, 24,037-- 24,046. Yeh, K. C., and C. H. Liu (1982), Radio wave scintillations in the ionosphere, Proc. IEEE, 70(4), 325-- 378. E. Bonelli, Departmento de Física, Universidade Federal do Rio Grande do Norte, Natal, RN 59072-970, Brazil. (bonelli@ponta-negra. com) 9of9