Improving terrestrial magnetosphere models with optimized ionospheric boundary conditions

Josh Rigler

Boundary conditions required for global terrestrial magnetohydrodynamic (MHD) simulations typically consist of in situ solar wind measurements (outer boundary), and plasma velocities determined by models of ionospheric electrical potential (inner boundary). Due to the large spatial scales and high speeds (millions of km and km/h) found in the solar wind, point measurements of the solar wind made upstream of the Earth are assumed to be valid for all outer boundary positions, given a simple ballistic correction.

The inner boundary is more problematic since it is characterized by much smaller spatial scales exhibiting substantial dynamic variability, and it is very poorly sampled. Semi-empirical relationships are used to transform standard MHD state variables into initial electron flux and energy values at the MHD model inner boundary. These are then mapped along static magnetic field lines to ionospheric altitudes, where they "precipitate" out to influence ionospheric conductances. These in turn impact electrical potential patterns, which are mapped back up the field lines to the inner MHD boundary to determine the MHD plasma velocity.

The Lyon-Fedder-Mobarry (LFM) global MHD model employed by HAO's Atomosphere-Ionosphere-Magnetosphere (AIM) group uses three adjustable parameters to define the electron energy and flux values required by ionospheric models that define its inner boundary conditions. These were tuned by hand many years ago so as to match in situ satellite data measured during relatively benign magnetospheric conditions. As a result, the LFM model has long been known to under-estimate electron energies and fluxes during so-called magnetospheric storms, active periods characterized by enhanced electron precipitation into the ionosphere that are driven by extreme variations in the solar wind, and ultimately the structure and dynamics of the solar corona.

We have taken some preliminary steps toward re-optimizing these three adjustable model parameters, including a nonlinear least-squares fit to remote sensing data that provide a more global view of aurora, the primary visible manifestation of electrons precipitating into the ionosphere during magnetospheric storms. We were able to improve the match between observations and predictions on average, but high spatial and temporal resolution features are missed almost entirely. What's more, there are indications that the re-optimized parameters no longer satisfy certain physical assumptions about the system being modeled. It is unclear at this point whether these problems arise from inherent limitations of the LFM model, our optimization algorithm, the training data, or (most likely) a combination of all the above.

Back to agenda