The future of geomagnetic storm predictions: implications from recent solar and interplanetary observations

Within the last 7–8 years, there has been a substantial growth in our knowledge of the solar and interplanetary causes of geomagnetic storms at Earth. This review article will not attempt to cover all of the work done during this period. This can be found elsewhere. Our emphasis here will be on recent efforts that expose important, presently unanswered questions that must be addressed and solved before true predictability of storms can be possible. Hopefully, this article will encourage some readers to join this effort and perhaps make major contributions to the field.     

Modulation of the solar daily geomagnetic variation

A least squares spectral analysis is used to investigate cyclic and seasonal changes of the harmonic coefficients of the solar daily and semi-diurnal variation over the 24 year interval from 1 January 1960 to 31 December 1983 at a single magnetic observatory (Dourbes, Belgium). As a preparation for the treatment of other long runs of observations a statistical method is presented to combine the output of the spectral and harmonic analyses of a group of stations. The annual means of the Fourier coefficients are significantly correlated with the solar cycle. The spectra show peaks at periods of 11 years, 1 year, 6 months and 27 days, which entails an important amplitude and phase modulation of the daily and semi-diurnal variation. No simple relationship between the peaks in the broad solar rotation band can be proposed.     

The geomagnetic time and position of a terrestrial station

The concepts of geomagnetic time and position of a station on Earth are reviewed on the basis of the centred dipole, and the eccentric dipole, magnetic field approximations. Equations are presented from which the centred dipole coordinates can be computed as a special case of the eccentric dipole coordinates. It is also shown that several aspects relating to geomagnetism appear to vary nearly linearly with time and are well represented by linear regression equations which can significantly simplify subsequent computations. Sufficient information is included so that magnetic time and position for any station on Earth may be calculated. Sample results are presented which specifically relate to Scott Base, Antarctica.     

The impact of gaps and spectral methods on the spectral slope of the middle atmospheric wind

The spectral slope of the middle atmospheric wind is an important index of the gravity wave and turbulence processes. Gaps exist in MF radar spaced-antenna winds data because significance criteria are built into the analysis. These cause a smearing of the spectrum and seriously modify the slope, as well as affect the absolute power at high frequencies. A comparison between sites with different gap rates must account for this. Different methods of dealing with these gaps are tested in this paper. The periodogram (with linear interpolation across gaps), the correlogram, and the Lomb-Scargle analyses are compared on synthetic data with known slope, and also with some of the best measured data (less than 20% gaps), both with added gaps to a maximum of 50%. The periodogram is seen to be the best choice. Parallel calculations on real data and synthetic data with the real gaps inserted are used to compare 1992 summer and winter spectral slopes from the Saskatoon MF radar. The latter are also compared with those of winter spectra from the two CNSR (Canadian Network for Space Research) radars which, with Saskatoon, form a ∼ 500 km array. A similar process is used to compare the seasonal variation of absolute power (10–100 min) at the three sites.     

Dependence of ionospheric response on the local time of sudden commencement and the intensity of geomagnetic storms

A study has been designed specifically to investigate the dependence of the ionospheric response on the time of occurrence of sudden commencement (SC) and the intensity of the magnetic storms for a low- and a mid-latitude station by considering total electron content and peak electron density data for more than 60 SC-type geomagnetic storms. The nature of the response, whether positive or negative, is found to be determined largely by the local time of SC, although there is a local time shift of about six hours between low- and mid-latitudes. The time delays associated with the positive responses are low for daytime SCs and high for night-time SCs, whereas the opposite applies for negative responses. The time delays are significantly shorter for mid-latitudes than for low-latitudes and, at both latitudes, are inversely related to the intensity of the storm. There is a positive correlation between the intensity of the ionospheric response and that of the magnetic storm, the correlation being greater at mid-latitudes. The results are discussed in the light of the possible processes which might contribute to the storm-associated ionospheric variations.     

The sensitivity of stratospheric photodissociation rates to the solar spectral resolution in the Schumann-Runge bands

Several molecules important in the chemistry of the stratosphere, including N₂O, H₂O, HNO₃, and several chlorofluorocarbons photodissociate in the wavelength interval 1750–2050 Å. The transmission of solar radiation in this spectral region is largely controlled by the Schumann-Runge (S-R) bands of molecular oxygen. The absorption cross sections of oxygen in this region vary by several orders of magnitude as a function of wavelength, resulting in large variations in the magnitude of solar radiation penetrating into the middle atmosphere. It is thus of interest to examine the potential effects that a knowledge of the fine structure of the solar spectrum would have on the accurate calculation of stratospheric photodissociation rates. A U.S. Naval Research Laboratory solar spectrum with a resolution of 0.07 Å has been used to compute high resolution photodissociation rates for several molecular species. Computations have also been performed in which this solar spectrum has been degraded by roughly a factor of 100. These calculations show that knowledge of the fine structure in the S-R region of the solar spectrum produces only a marginal (1–2%) improvement in the computation of the photodissociation rates. This results from the fact that the photodissociation rate is a spectrally integrated quantity, and that there is, on average over the S-R bands, no correlation between the solar spectrum and the oxygen absorption cross sections. This very small improvement should be compared with the much larger uncertainties in important chemical reaction rates, absorption cross sections, and quantum yields, which are often 50–100%.     

Lunar and solar daily variations of the geomagnetic field at Italian stations

Mean hourly values of magnetic declination D, horizontal intensity H and vertical intensity Z observed at Italian stations have been analyzed to determine solar and luni-solar diurnal components, together with the corresponding terms O₁ and N₂ of the lunar tidal potential.The results, showing the variations of the first four harmonic components with season, degree of magnetic activity and annual sunspot number, are tabulated and discussed. Differences between the dependence of S and L on season and sunspot number are considered and tentative explanations offered. The oceanic tidal effect has been determined and it is apparent that this is more likely to show the influence of the Atlantic Ocean rather than the Mediterranean Sea.     

Upper atmosphere wind observations over Southern Australia during the total solar eclipse of 23 October 1976

Observations of winds in the 80–100 km height region were made at three locations in South Australia during the total solar eclipse of 23 October 1976. One station (Tantanoola) was located in the eclipse path while the others (Adelaide and Woomera) were situated several hundred kilometers north of the path of totality. Wind variations caused either directly by cooling of the 90 km region or by the propagation of a bow wave generated in the lower atmosphere were searched for but no events were found that could be ascribed unambiguously to the eclipse.     

The 27 day solar UV response of stratospheric ozone: solar cycle 21 vs. solar cycle 22

A correlative study of ozone and the solar UV flux on the time scale of a solar rotation shows an anomalous response of ozone in the upper stratosphere during solar cycle 22. The study, which is based on the analysis of ozone and solar UV flux measured by the SBUV/2 spectrometer on NOAA 11 (January 1989–December 1990), shows a sharp transition from an in-phase relation between ozone and the solar UV flux below 2 mb to an almost out-of-phase relation above 1 mb. Such a phase change is not predicted by photochemical models and was not observed during solar cycle 21. The ozone measurements from the Nimbus-7 SBUV spectrometer from 1979 to 1984 showed an almost in-phase relation between ozone and the solar UV flux at these heights (in agreement with model predictions). Similar studies of ozone and temperature relations between 30 and 1 mb did not show significant changes from the solar cycle 21 to 22. The temperature oscillations appear to be primarily of dynamical origin, with no apparent correlation with solar UV flux.     

Penetration of solar wind plasma elements into the magnetosphere

Assuming that the solar wind is unsteady and non-uniform, it is suggested that field aligned plasma elements dent the magnetopause surface. This indentation makes the magnetopause boundary convex and therefore locally unstable with respect to flute instabilities. The intruding element is slowed down and stopped within 1 or 2 Earth radii from the magnetopause. Its excess convection kinetic energy is dissipated in the lower polar cusp ionosphere after ~ 50–500 s, depending on the value of the integrated Pedersen conductivity. Once the plasma element has been engulfed, keeping its identity, the warm plasma content is dissipated by precipitation and by drifting. The magnetosheath particles with large pitch angles are mirrored, and feed the plasma mantle flow.Several consequences of this penetration mechanism are pointed out:Ionospheric heating beneath the polar cusp;Birkeland currents on the eastward and westward edges of the plasma element;Diamagnetic field fluctuations within 1–2 R_E from the magnetopause (multiple magnetopause crossings);Oscillation of the magnetopause surface after a new element has penetrated;Exit of energetic particles out of the magnetosphere, and entry of energetic solar wind particles into the magnetosphere along the magnetic field lines of the intruding element;Subtraction of magnetic flux from the dayside magnetosphere and its addition to the geomagnetic tail when the magnetic field of the element has a southward component.     

Morphology of low-frequency waves in the solar wind and their relation to ground pulsations

Three classes of low frequency waves (period range 20–80 s) were identified using data from the UCLA fluxgate magnetometer experiment on board the ISEE 2 spacecraft. These are continuous pulsations similar in type to Pc 3, band-limited oscillations distinguished by mixed period fluctuations, and relatively isolated wave bundles. The waves were preferentially observed when the interplanetary magnetic field (IMF) direction was sunward and were most common when the cone angle, i.e. the angle between IMF and the Sun-Earth line (θ_xb) was often between 15° and 45°. Their frequency is proportional to the IMF magnitude.Comparison between the waves observed on board the ISEE 2 spacecraft and the Pc 3–4 recorded simultaneously at a mid-latitude ground station, Oulu (L = 4.5), showed that similarity of spectra of the waves in the spacecraft and on the ground was very rare and that correspondence between the events in space and on the ground was extremely low.     

The trouble with thermospheric vertical winds: geomagnetic,seasonal and solar cycle dependence at high latitudes

The vertical wind component is frequently used to determine the zero-velocity baseline for measurements of thermospheric winds by Fabry-Perot and other interferometers. For many of the upper atmospheric emission lines from which Doppler shifts are determined, for example for the OI 630 nm emission, available laboratory sources are not convenient for long-term use at remote automatic observatories. Therefore, the assumption that the long-term average vertical wind is zero is frequently used to create a baseline from which the Doppler shifts corresponding with the line-of-sight wind from other observing directions can then be calculated. A data base consisting of 1242 nights of thermospheric wind measurements from Kiruna (68°N, 20°E), a high-latitude site, has been analysed. There are many interesting short-term fluctuations of the vertical wind which will be discussed in future papers. However, the mean vertical wind at Kiruna also has a systematic variation dependent on geomagnetic activity, season and solar cycle. This means that the assumption that the average value of the vertical wind is zero over the observing period cannot be used in isolation to determine the instrument reference or baseline. Despite this note of caution, even within the auroral oval, the assumption of a zero mean vertical wind can be used to derive a baseline which is probably valid within 5 ms⁻¹ during periods of quiet geomagnetic activity (K_p < 2), near winter solstice. During other seasons, and during periods of elevated geomagnetic activity, a systematic error in excess of 10 ms⁻¹ may occur.     

Simulation of ionospheric currents and geomagnetic field variations of Sq for different solar activity

Variations of ionospheric Sq electric currents and fields caused by changes in electric conductivity due to changes in solar activity are studied using the International Reference Ionosphere (IRI) model. Calculations are made for R (sunspot number) = 35 and 200 on the assumption of constant (1, −2)mode tidal winds. It is shown that electric fields grow when solar activity is high, because the ratio of the conductivity in the F-region to that in the E-region increases. Currents in the F-region become stronger than those in the E-region, and nocturnal currents are not negligible when solar activity becomes high. F-region currents also play an important role in the westward currents on the high latitude side of the current vortex. The calculated geomagneticH component at the equator has a depression around 1600 LT for R = 35, while it decreases smoothly from 1100 LT to 1900 LT for R = 200. This difference is consistent with the observed geomagnetic field variation. The ratio of total Sq currents obtained by our simulation is about 3.5, which is a little larger than is found in the observed results.     

Positive and negative ion composition measurements in the D- and E-regions during the 26 February 1979 solar eclipse

Two rockets bearing quadrupole mass spectrometers capable of measuring both positive and negative ion composition were launched from Red Lake, Canada, during the solar eclipse. Both instruments had liquid helium cryopumps and shock-attaching conical samplers. The payloads also contained two Gerdien condensers to measure total positive and negative ion concentrations and ion mobilities. Attitude control systems aligned the payloads with the velocity vector throughout ascent and descent. The first rocket was launched so that the D-region was in darkness 35 ± 8 s on the upleg and about 150 ± 15 s on the downleg for the study of ionospheric decay processes. The second rocket was fired after totality into 75% solar illumination for the study of ionospheric recovery. The positive ion composition above 105 km exhibited a strongly increasing NO⁺/O₂⁺ ratio with time after second contact due to O₂⁺ charge transfer with NO and a sharply diminished ionization rate. However, in both nights, the ionization below 105 km was created mainly by energetic particle deposition as exemplified by the increased ion concentrations and the composition signatures of a particle event: asignificant enhancement of O₂⁺ below 105 km and large amounts of H₅O₂⁺ ions in the D-region which result from the O₂⁺ clustering scheme. H₅O₂ was the major ion in the upper D-region while H₇O⁺₃, H₉O₄⁺ and H₅O₂⁺ were dominant ions at lower altitudes. Numerous minor species were also detected. The negative ion distributions in both flights exhibited a distinct shelf at 83 ± 2 km, decreasing by more than an order of magnitude by 90 km and with minima near 75 km. In the 75–90 km range, a significant percentage of the negative ions had masses exceeding 160 a.m.u. Comparisons are made with prior negative ion measurements during similar daytime auroral zone absorption (AZA) events. Two striking characteristics of the precipitating particles were apparent from these and past observations in daytime AZA events: there is a near absence of low energy electrons capable of ionizing above about 105 km and there is'a significant spatial and/or temporal variability in the electron flux. This paper is devoted principally to a presentation of the ion composition measurements and associated uncertainties.     

Analysis of the connections between solar wind parameters and day-side geomagnetic pulsations based on data from the observatories Nagycenk (Hungary) and Uzur (U.S.S.R.)

A comparison is made of results obtained at two widely separated observatories, Nagycenk in Hungary and Uzur in Soviet Siberia (both at geomagnetic latitude 48°) and the connection is studied between pulsation parameters and parameters of the interplanetary medium. It is concluded that, in spite of differences in recording and processing, the results are similar and confirm earlier suppositions about these connections. Hypotheses about the location of the excitation clearly favour the pre-shock interplanetary medium with reflected protons counterstreaming to the solar wind. Propagation through the shock wave and the magnetopause, as well as infra-magnetospheric propagation, may later alter certain parameters of the pulsations.     

Interaction of solar wind with the magnetopause-boundary layer and generation of magnetic impulse events

The transport of mass, momentum, energy and waves from the solar wind to the Earth's magnetosphere takes place in the magnetopause-boundary layer region. Various plasma processes that may occur in this region have been proposed and studied. In this paper, we present a brief review of the plasma processes in the dayside magnetopause-boundary layer. These processes include

• 1.(1) flux transfer events at the dayside magnetopause,
• 2.(2) formation of plasma vortices in the low-latitude boundary layer by the Kelvin-Helmholtz instability and coupling to the polar ionosphere,
• 3.(3) the response of the magnetopause to the solar wind dynamic pressure pulses and
• 4.(4) the impulsive penetration of solar wind plasma filaments through the dayside magnetopause into the magnetospheric boundary layer. Through the coupling of the magnetopause-boundary layer to the polar ionosphere, those above processes may lead to occurrence of magnetic impulse events observed in the high-latitude stations.

     

Long-period variations of wind parameters in the mesopause region and the solar cycle dependence

A solar dependence of wind parameters below 100 km was found by Sprenger and Schminder on the basis of long-term continuous ionospheric drift measurements (D1) in the l.f. range. For winter they obtained for the prevailing wind a positive correlation with solar activity and for the amplitude of the semi-diurnal tidal wind a negative correlation. Later on this result was confirmed by radar meteor wind measurements (D2) at Obninsk and further D1 measurements at KÜhlungsborn and Collm.However, after the years 1973–1974 a change in the behaviour of the zonal prevailing wind was observed. At this time we found a significant negative correlation with solar activity with an indication of a new change after 1983. This was obtained from D1 results in Collm and D2 results in Kühlungsborn not only for winter, but also for summer and even for annual averages. We conclude that this long-term behaviour points rather to a climatic variation with an internal atmospheric cause than to a direct solar control. The negative correlation with solar activity of the semi-diurnal tidal wind in winter remained unchanged (up to 1984) and also proved to be the same in summer and for annual averages. Recent satellite data of the solar u.v. radiation and the upper stratospheric ozone have shown that the possible variation of the thermal tidal excitation during the solar cycle amounts to only a few per cent. This is, therefore, insufficient to account for the 40–70% variation of the tidal amplitudes. Some other possibilities of explaining this result are discussed.     

Application of refractive scintillation theory to radio transmission through the ionosphere and the solar wind,and to reflection from a rough ocean

The results of Booker and Majidiahi (1981) concerning refractive scattering by large-scale irregularities in a phase-changing screen are combined with the theory of diffractive scattering by small-scale irregularities in order to study three intensity scintillation phenomena. The first is the reflection of radio and optical waves from an ocean surface disturbed by a spectrum of water waves. The second is the scintillation of VHP, UHF and SHF radio waves traversing the ionospheric F-region. The third is the scintillation of VHF, UHF and SHF radio waves traversing the solar wind. In each case appropriate values are chosen for the mean square fluctuation of phase, for the outer scale, for the inner scale and for the spectral index. Spectral diagrams are drawn to show how the outer scale, the inner scale, the Fresnel scale, the focal scale, the lens scale and the peak scale vary with a relevant parameter (electromagnetic wave-frequency for the ocean, RMS fractional fluctuation of ionization density for the ionosphere, and distance of closest approach to the Sun for the solar wind). For the ionosphere and the solar wind, multiple refractive scattering by weak irregularities occurs in practice whereas it is strong single scattering that is assumed in the thin-screen theory ; potential consequences of this are discussed qualitatively.     

Changes of pulsation activity during two solar cycles

Monthly and yearly averages of the pulsation activity in the mid-latitude station Nagycenk are compared to solar wind velocity and ionospheric-plasmaspheric electron concentration data. It is found that pulsation amplitudes are correlated with solar wind velocities with the exception of some month around December in solar maximum years, when they are significantly lower than computed from the corresponding solar wind velocities. This decrease can be caused either by a cutoff of the magnetospheric shell resonances or by local ionospheric damping. In addition to these effects, pulsations amplitudes slightly depend also on the geomagnetic activity and have a semi-annual activity change with maxima around equinoxes.