The NCAR Thermosphere-Ionosphere-Electrodynamics General Circulation Model: Problems in Developing a Realistic Model

Astrid Maute, Art Richmond, and Ben Foster
NCAR/High Altitude Observatory Division

Our group has developed a simulation model of the upper atmosphere, the NCAR Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE- GCM; Dickinson et al., 1984; Roble et al., 1988; Richmond et al., 1992), which includes the primary physical and chemical processes that determine its structure and dynamics. Simulation results depend on the model inputs and boundary conditions,including solar radiation, magnetospheric currents, auroral particle fluxes, and global-scale atmospheric waves from below, but there are large uncertainties in what these highly variable inputs should be at any given time. Observations are relatively sparse, and extrapolation of the observations in space in time to get complete specification of the boundary conditions is uncertain. There are also uncertainties in some of the model parameters and parameterizations. For example, turbulent (eddy) diffusion has a big influence, but its magnitude and variability are poorly known. The rates of some chemical reactions and molecular collision processes are also uncertain, and there are large uncertainties in how to parameterize sub-grid-scale processes like atmospheric gravity waves. These uncertainties create difficulties for trying to simulate realistic upper-atmospheric behavior, both for average conditions and for events like magnetic-storm disturbances.

We will present an overview about the NCAR Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM) which should help to better understand how the model is working, and what physics is described. To point out the variability and the non-linearity we will show examples of different input parameters and their influence on the model output. We will also demonstrate the effect of uncertain model parameters and parameterization. One of the challenges in tuning a model is that observations are sparse in time and space, and that there is no formalism to get optimized model parameters based on the observations. We will show examples of observations to point out the difficulties.

For the SAMSI project we focused on three model parameters which were varied, and will be described. We will talk about their expected influence on the model results. In this first step we focused only on a few model outputs for which we also had observations: magnetic perturbations and the drift velocities.

Future work can include to extend the study to vary more model parameters, and work with more model output. Finally, we will point out where advanced statistical approaches might be able to help improve the models.

Back to agenda