The global ionosphere-thermosphere model

Abstract The recently created global ionosphere-thermosphere model (GITM) is presented. GITM uses a three dimensional spherical grid that can be stretched in both latitude and altitude, while having a fixed resolution in longitude. GITM is nontraditional in that it does not use a pressure-based coordinate system. Instead it uses an altitude-based grid and does not assume a hydrostatic solution. This allows the model to more realistically capture physics in the high-latitude region, where auroral heating is prevalent. The code can be run in a one dimensional (1-D) or three dimensional (3-D) mode. In 3-D mode, the modeling region is broken into blocks of equal size for parallelization. In 1-D mode, a single latitude and longitude is modeled by neglecting any horizontal transport or gradients, except in the ionospheric potential. GITM includes a modern advection solver and realistic source terms for the continuity, momentum, and energy equations. Each neutral species has a separate vertical velocity, with coupling of the velocities through a frictional term. The ion momentum equation is solved for assuming steady-state, taking into account the pressure, gravity, neutral winds, and external electric fields. GITM is an extremely flexible code -allowing different models of high-latitude electric fields, auroral particle precipitation, solar EUV inputs, and particle energy deposition to be used. The magnetic field can can be represented by an ideal dipole magnetic field or a realistic APEX magnetic field. Many of the source terms can be controlled (switched on and off, or values set) by an easily readable input file. The initial state can be set in three different ways: (1) using an ideal atmosphere, where the user inputs the densities and temperature at the bottom of the atmosphere; (2) using MSIS and IRI; and (3) restarting from a previous run. A 3-D equinox run and a 3-D northern summer solstice run are presented. These simulations are compared with MSIS and IRI to show that the large-scale features are reproduced within the code. We conduct a second equinox simulation with different initial conditions to show that the runs converge after approximately 1.5 days. Additionally, a 1-D simulation is presented to show that GITM works in 1-D and that the dynamics are what is expected for such a model