The Response of the Thermosphere and Ionosphere to Magnetospheric Forcing

During the past six years, rapid advances in three observational techniques (ground-based radars, optical interferometers and satellite-borne instruments) have provided a means of observing a wide range of spectacular interactions between the coupled magnetosphere, ionosphere and thermosphere system. Perhaps the most fundamental gain has come from the combined data-sets from the NASA Dynamics Explorer (DE) Satellites. These have unambiguously described the global nature of thermospheric flows, and their response to magnetospheric forcing. The DE spacecraft have also described, at the same time, the magnetospheric particle precipitation and convective electric fields which force the polar thermosphere and ionosphere. The response of the thermosphere to magnetospheric forcing is far more complex than merely the rare excitation of 1 km s^-1 wind speeds and strong heating; the heating causes large-scale convection and advection within the thermosphere. These large winds grossly change the compositional structure of the upper thermosphere at high and middle latitudes during major geomagnetic disturbances. Some of the major seasonal and geomagnetic storm-related anomalies of the ionosphere are directly attributable to the gross wind-induced changes of thermospheric composition; the mid-latitude ionospheric storm `negative phase', however, is yet to be fully understood. The combination of very strong polar wind velocities and rapid plasma convection forced by magnetospheric electric fields strongly and rapidly modify F-region plasma distributions generated by the combination of local solar and auroral ionization sources. Until recently, however, it has been difficult to interpret the observed complex spatial and time-dependent structures and motions of the thermosphere and ionosphere because of their strong and nonlinear coupling. It has recently been possible to complete a numerical and computational merging of the University College London (UCL) global thermospheric model and the Sheffield University ionospheric model. This has produced a self-consistent coupled thermospheric--ionospheric model, which has become a valuable diagnostic tool for examining thermospheric--ionospheric interactions in the polar regions. In particular, it is possible to examine the effects of induced winds, ion transport, and the seasonal and diurnal U.T. variations of solar heating and photoionization within the polar regions. Polar and high-latitude plasma density structure at F-region altitudes can be seen to be strongly controlled by U.T., and by season, even for constant solar and geomagnetic activity. In the winter, the F-region polar plasma density is generally dominated by the effects of transport of plasma from the dayside (sunlit cusp). In the summer polar region, however, an increase in the proportion of molecular to atomic species, created by the global seasonal circulation and augmented by the geomagnetic forcing, controls the plasma composition and generally depresses plasma densities at all U.Ts. A number of these complex effects can be seen in data obtained from ground-based radars, Fabry--Perot interferometers and in the combined DE data-sets. Several of these observations will be used, in combination with simulations using the UCL--Sheffield coupled model, to illustrate the major features of large-scale thermosphere--ionosphere interactions in response to geomagnetic forcing. The past decade has seen a major improvement in the quality and quantity of experimental data available to study the thermosphere and ionosphere and their response to magnetospheric forcing. Earlier, large measured changes of individual parameters were difficult to place in a global or large-scale perspective. However, a clear picture of the distinction between the solar and geomagnetic forcing processes has emerged from the combined data-sets available from spacecraft such as the Dynamics Explorers, and from ground-based radar and optical observations of the polar thermosphere. A first experimental view of the strong coupling between the thermosphere and ionosphere has also emerged from these combined new data-sets. In parallel with the development of observing techniques, numerical models of the thermosphere and ionosphere have matured. We are at a state where the combined thermosphere and ionosphere can be modelled self-consistently. We can now realistically simulate the response of the combined system to the magnetospheric forcing, and also investigate the many and varied feedback processes between the two components. The models can be used to understand and interpret the diversity of experimental observations, and provide the framework for evaluating phenomena which are as yet not well understood. The dominant thermosphere--ionosphere interactions which appear from the modelling studies and which have counterparts in the experimental database can be summarized. In the winter polar region, ionization enhancements are observed which are due to auroral particle precipitation in both the E-region and in the F-region. The former are relatively easy to understand, since decay rates are generally rapid, and large-scale transport is unimportant. The sole caveat will be related to sporadic-E layers of long-lived metallic ions. In the polar F-region, neutral winds, neutral composition changes, convection changes and solar photoionization all cause important modifications of plasma distributions. In the winter, plasma convection and winds cause important effects in the horizontal and vertical transport of plasma, respectively. As such, plumes of high density (or low-density) plasma are transported large distances from their origin, and local plasma densities are rarely explicable by local sources and sinks. The exact distributions will depend very much on detailed plasma convection patterns. However, the winter subauroral trough and localized polar troughs will be created when the combination of convection and corotation cause plasma stagnation in regions out of sunlight and photoionization. There is a strong U.T. modulation of plasma density within the winter polar cap and dusk auroral oval (generally) as the polar cusp enters sunlight for a few hours around 18h U.T., and there is a direct source of high-density plasma (photoionization plus particle ionization) convected through the cusp. At other U.Ts, the source is generally cut off, and polar plasma densities generally decay. Summer F-region high-latitude and polar plasma densities are generally a factor of about 3-5 lower than in winter. This is due to the seasonal F-region neutral composition variation, generated by summer to winter mean circulation, by which increased plasma recombination rates (due to much higher molecular nitrogen densities) more than compensate for the increased solar photoionization source in the summer polar cap. In turn, this mean circulation is generated by the combination of asymmetric solar insolation and greater geomagnetic heating in the summer compared with winter polar regions (Rees et al. 1985, 1987). Particularly at times of high geomagnetic activity, the summer `F-region' neutral composition is close to that of the standard atmosphere E-region. The major features of the summer polar F-region are thus quite different to those of the corresponding winter region. Plasma troughs develop in regions of very strong Joule heating, i.e. where ion convection is strongest. As such, the location and intensity of the troughs is quite dependent on the plasma convection patterns. Summer-time troughs tend to occur in the same regions where rapid transport causes high-density plasma plumes in the winter polar region. The classical subauroral trough is distinctly a feature of the winter polar F-region. Even at equinox, the full subauroral trough does not develop, while in winter it fully encircles the geomagnetic polar cap for much of the U.T. day (except around 18h U.T.). In the summer F-region, stagnation troughs do not develop within the polar cap, irrespective of convection pattern. Any polar cap troughs are a result of changes in neutral composition. Subauroral troughs can only develop around the summer polar region when the auroral oval is expanded so that the midnight part of the auroral oval extends into the nightside. Conditions for this situation are likely to occur preferentially in the southern polar region, due to the greater offset of the geomagnetic from geographic pole. The E-region response to geomagnetic forcing is also strong, although generally rather less marked than in the F-region, in terms of the neutral thermal and compositional response. The major feed-back between the thermosphere and ionosphere occurs due to the effects of high induced winds, since the neutral chemical changes do not significantly affect the ionospheric chemistry. Apart from localized effects such as sporadic-E layers, high-speed auroral oval winds do not cause significant vertical transport of E-region molecular species, due to rapid recombination. The most significant vertical transport effects will be in non-sunlit regions, where ion production is lowest. The dynamo effect of induced E-region winds of 200-400 m s^-1 is also quite significant. Such winds reduce horizontal currents, with an implication that the FAC or Pedersen currents may also be decreased, with a possible feedback to the convection electric field. There is still relatively little data available for detailed case-study comparisons. Such studies have been quite successful in improving our understanding of the F-region behaviour, and the CEDAR initiative and programmes such as LTCS promise to extend the range of multiparameter data-sets to the E-region as well. Simulating atmospheric density and compositional structure with numerical models is one of the most testing demands. Density at a given altitude is very sensitive to the total thermospheric energy budget, and is thus liable to be the first casualty of cumulative small errors in the many external terms of the energy input. There are also some indeterminate factors in the radiative energy budget of the lower thermosphere and upper mesosphere. In practice, we have found that the present version of the coupled model computes density and composition relatively accurately, compared with mean mass spectrometer and incoherent scatter (MSIS) predictions for comparable solar and geomagnetic activity levels and for different seasonal conditions. Typical differences (MSIS to model) of around 20% occur at F-region altitudes in the data-sets shown in the model simulations described within this paper. This is roughly comparable with the standard deviation of MSIS in comparison with satellite data-sets for specific locations and times. The numerical models have greater spatial and temporal resolution than MSIS models and relate to real physical processes. Undoubtedly, however, the real thermosphere contains a whole spectrum of high-frequency variations which are beyond present parametrization techniques, our current description of geomagnetic inputs and present computer limitations. From the initial coupled-model simulations it is possible to examine key features of the coupling between the magnetosphere and the thermosphere--ionosphere. Field-aligned currents reflect the divergence or convergence of the ionospheric Pedersen current. The Pedersen current depends on changes of the ionospheric conductivity and also the dynamo effects of induced winds. Both FAC and E-region winds display considerable seasonal, U.T. and geomagnetic activity variations. Except in the unlikely event that the magnetosphere acts as a `zero-resistance' source of charge, and momentum, etc., we would anticipate, on the basis of these thermosphere--ionosphere model simulations, to see corresponding modulation of magnetosphere--ionosphere forcing as a function of U.T., season and geomagnetic activity. However, a detailed theoretical evaluation of such processes will have to await the development of a new range of coupled models embracing the near-Earth environment. As new experimental data from coordinated ground-based campaigns becomes available over the next several years, and it is to be hoped from new space missions within the next decade, we may hope that the validity of many of the simplified assumptions we currently have to make within present models can be tested. Undoubtedly, many present concepts will be found wanting. The impact of global images of particle precipitation and energy deposition, coupled with perhaps the development of techniques of imaging polar plasma convection patterns will mean that future models are capable of looking at the effects of short period and smaller-scale variations in forcing. The present patterns of magnetospheric forcing are too simplified and averaged in time and space. While the thermosphere averages out rapid and short-scale momentum inputs, the energy input integrates all variations, including the effect of rapid forcing variations. The thermospheric composition responds to this `additional' energy source in a way which presently cannot be simulated accurately, and we already know how sensitive the polar plasma environment appears to be to thermospheric composition changes forced by the combined solar and magnetospheric forcing. We are indebted to Dr Fred Rich for provision of the Heppner & Maynard polar electric fields in the form of harmonic coefficients. We also thank John Harmer and Hilary Hughes for their assistance in preparing, running and processing the computer simulations using the UCL--Sheffield coupled ionospheric--thermospheric model. Computer time was made available by the University of London Computer Centre (CRAY 1-S) and on the CRAY-XMP-48 at the Rutherford Appleton Laboratory (Science and Engineering Research Council). The research was supported by grants from the U.K. SERC, and from the European Office of Aerospace Research and Development (AFOSR-86-341). The IGRF magnetic field model was provided, in computer-readable form, by the British Geological Survey, Edinburgh.