general, the model has been successful in reproducing electrodynamics in the high‐latitude ionosphere (Lu et al., 1995; Ridley et al., 1997, and many others). As in the case of the W05 and Cosgrove14 models, AMIE has been used widely as input to the GCMs. Unlike the empirical models, AMIE captures near‐real‐time variations in high‐latitude electrodynamics, with some smoothing in the assimilation scheme. Temporal and spatial resolution depend on the data to be assimilated.
However, there at two caveats. The assumption that the currents in the ionosphere can be represented by equivalent currents on the ground needs to be applied with caution. The assumption is valid in the case of Hall currents, which are detected by ground magnetometers. Comparisons of satellite‐based measurements from CHAMP and ground detections of currents from the IMAGE network have validated the equivalent current assumption (Ritter et al., 2004). However, ground‐detection of Pedersen current is inhibited by the shielding effect of FACs (Araki et al., 1989). As a result, correlation of ground‐ and space‐based observations of Pedersen currents is poor (Ritter et al., 2004). In practice, AMIE uses an empirical model to relate the measured Hall current to Pedersen current (Ahn et al., 1983a). But the empirical models are frequently based on observations limited to the auroral regions. Application of the results to subauroral or polar latitudes is questionable.
The only region at high latitudes in which Hall current dominates is the auroral zone in which energetic particle precipitation is the norm (Newell et al., 2005, and references therein). At latitudes poleward and equatorward of the auroral zone, or Boundary Plasma Sheet as described by Newell et al. (1991), which are characterized by soft electron precipitation, the use of empirical models to estimate Pedersen conductivity may distort the true picture. The lack of ground signature of ionospheric electrodynamics was illustrated in a study of a superstorm on 6 April 2000 by Huang and Burke (2004). During the storm, four DMSP spacecraft detected intense FACs, with ΔB between 1,000 and 1,400 nT at 840 km altitude, at latitudes equatorward of 60° MLat. At the conjugate location on the ground, the maximum magnetic perturbation was under 50 nT.
The second cautionary note on application of AMIE is related to the first, and derives from the conductance that is required for the synthesis of observations. The standard model referenced for use with AMIE is by Ahn et al. (1983a). The estimate of conductivity in this model assumes energetic electron precipitation leading to Hall current at altitudes between 100 and 125 km. Pedersen conductivity is estimated from Hall conductivity as a ratio derived empirically. Robinson et al. (1987) give a similar relationship. However, these empirical ratios are derived for auroral precipitation in which electron energies are generally assumed to be ≥1 keV. Huang and Burke (2004) estimated the Pedersen conductance using measured magnetic deflections and precipitating electron fluxes from DMSP F13 and F15 satellites for the superstorm on 6 April 2000. The relation between conductance, magnetic field, and electric field is simply Ohm’s law, J = ∑ x E, where J and E are the height‐integrated current density and electric field, respectively. Figure 1.6 compares the values obtained from the DMSP data with the Robinson et al. (1987) formulas for Pedersen and Hall conductances. The horizontal bars indicate the time intervals when Pedersen conductance can be estimated directly. It is clear that the empirical results (Ahn et al., 1983a; Robinson et al., 1987) do not represent the conductance well in regions where keV electrons are not dominant.
Figure 1.6 (a)–(b) Values of Pedersen (∑P) and (c)–(d) Hall (∑H) conductance estimated from energetic electron fluxes measured by (a) and (c) DMSP F15 and (b) and (d) F13 using the formula by Robinson et al. (1987) during a superstorm on 6 April 2000. F15 is located at ‐52° MLat, 221 MLT. F13 is located at 59.9° MLat, 6.7 MLT. Heavy horizontal lines indicate values of ∑P estimated from measured δBz and EY variations
(figure and caption based on Huang & Burke, 2004. Reproduced with permission of John Wiley & Sons).
Given our understanding that EM energy dominates the energy budget, the assumptions underlying the model should be examined for relevance. In particular, the role of the auroral zone as the locus of all energy input has been questioned (Huang et al., 2014) based on the total energy budget during a magnetic storm. This may be addressed by AMIENextGen, an updated version of AMIE, which is being developed (Matsuo et al., 2015; Cousins et al., 2015; McGranaghan et al., 2016). In AMIENextGen, many of the data sources are based on satellite observations, which may avoid some of the difficulties with AMIE.
1.3 GENERAL CIRCULATION MODELS (GCMS) OF MIT COUPLING
The GCMs represented a significant advance in physics‐based MIT modeling, capturing the essential physics in the IT system. They include the chemical reactions dominant at low altitudes, and the coupled equations of momentum, energy, and continuity for ion, electrons, and neutrals (Rees et al., 1980; Fuller‐Rowell et al., 1988, 2000; Dickinson et al., 1981, 1984; Roble et al., 1987; Ridley et al., 2006). They have been used to simulate the ionospheric response to energy deposition and dissipation at high latitudes in many publications. We focus on three of the most widely used models: the Thermosphere Ionosphere Electrodynamics Global Circulation Model (TIEGCM), the Coupled Thermosphere Ionosphere Plasmasphere Electrodynamics (CTIPe) model, and the Global Ionosphere‐Thermosphere Model (GITM), all of which have been run for many years. Model runs on demand of all three are available at the Community Coordinated Modeling Center (CCMC) (http://ccmc.gsfc.nasa.gov/).
All three require specification of the high‐latitude electric field or high‐latitude Poynting flux and conductivity in order to simulate the effect of energy deposition and dissipation. The electric field or Poynting flux can be specified by W05, Cosgrove14, or AMIE as described in section 1.2. The conductivity can be obtained from empirical models by Roble and Ridley (1987) or Fuller‐Rowell and Evans (1987).
The CTIPe model is a nonlinear, coupled thermosphere‐ionosphere‐plasmasphere physics‐based code that includes a self‐consistent electrodynamics scheme for the computation of dynamo electric fields. There are four distinct components that run simultaneously and are coupled: a global thermosphere, a high‐latitude ionosphere, a mid‐and low‐latitude ionosphere/plasmasphere, and an electrodynamical calculation of the global dynamo electric field (Fuller‐Rowell & Rees, 1980; Codrescu et al., 2012). The thermospheric component is divided into a geographic latitude x longitude grid with resolution 2° x 18°. The vertical resolution is defined in terms of the logarithm of pressure from a lower boundary at 80 km altitude to altitudes above 500 km. The primary inputs to the model are the high‐latitude electric field, usually provided by the Weimer (2005) empirical model, auroral electron precipitation from the empirical model by Fuller‐Rowell and Evans (1987), and EUV radiation (Solomon & Qian, 2005). Full descriptions of the model with examples of output can be found in Codrescu et al. (2012). CTIPe is run in near‐real time at the Space Weather Prediction Center (SWPC), and model output is available for comparison with observations at http://helios.swpc.noaa.gov/ctipe/index.html.
TIEGCM provides a three‐dimensional, nonlinear representation of the coupled IT system. A self‐consistent calculation of ionospheric wind dynamic effects (Richmond et al., 1992) is included. The primary external drivers of the model are solar irradiance, magnetospheric energy, and tidal perturbations at the lower boundary of the model. Magnetospheric energy inputs include particle precipitation