On Thermospheric Density and Wind Modeling Driven by Satellite Observations

Abstract

The thermosphere is home to a plethora of orbiting objects ranging in size from flecks of paint to modular spacecraft with masses on the order of thousands of kilograms. The region spans hundreds of kilometers in vertical extent, from ∼100 km where fixed-wing flight by aerodynamic lift is unsupportable, out to ∼500-700 km, depending on solar activity, where the particle density is so sparse that the atmosphere can no longer be treated as a fluid. The thermosphere is subject to dynamical energy input from radiation and magnetic sources that make quantifying its dynamics a nontrivial endeavor. This is particularly a challenge during geomagnetic storms, where increased magnetic activity primarily at high-latitudes drives global heating, traveling atmospheric disturbances, and intense winds throughout the thermosphere. Modeling of the neutral density and horizontal winds is a challenging endeavor for these conditions, and it is vital not only for understanding the physics of neutral atmospheres, but also for the practical purposes of improving orbit prediction, as the thermosphere is home to an increasing number of satellite missions, in addition to being the abode of astronauts. Various atmospheric models have been constructed and developed over decades in order to model the thermosphere, with the most prominent being the empirical models Mass Spectrometer and Incoherent Scatter Radar MSIS-00, Jacchia-Bowman JB-2008, and Drag-Temperature Model DTM-2013, which are primarily used to model the neutral density, and GITM, a physics-based model capable of modeling atmospheric electrodynamics and investigating thermospheric winds. This dissertation focuses on three important means by which the interplay between satellite measurements and atmospheric models can drive scientific development for use in satellite mission operations and model development outright. In order to reduce the empirical mode bias during storms, we created the Multifaceted Optimization Algorithm (MOA), a method to modify the drivers of the models by comparing actual and simulated orbits through the model to reduce the errors. Applying MOA to the MSIS-00 model allowed a decrease in model error from 25% to 10% in the event that was examined, and represents an easy-to-implement technique that can use publicly available two-line-element orbital data. A superposed epoch analysis of three empirical density models shows persistent storm-time overestimation by JB-2008 and underestimation DTM-2013 by MSIS-00 for more intense geomagnetic storms that may be addressed with a Dst-based calibration, and a statistical analysis of GITM horizontal winds indicates the best performance in the polar and auroral zones and difficulty capturing seasonality. The work contained in this dissertation aims to provide techniques and analysis tools to improve density and wind model performance, in order to support satellite mission operators and atmospheric research. Ultimately, it demonstrates that simple tools and methods can be utilized to generate significant results and scientific insight, serving to augment and supplement more computationally intensive and cost-prohibitive strategies for investigating the thermospheric environment.