Spatial statistics and analysis of earth's ionosphere
Butler, Thomas W.
MetadataShow full item record
The ionosphere, a layer of Earths upper atmosphere characterized by energetic charged particles, serves as a natural plasma laboratory and supplies proxy diagnostics of space weather drivers in the magnetosphere and the solar wind. The ionosphere is a highly dynamic medium, and the spatial structure of observed features (such as auroral light emissions, charge density, temperature, etc.) is rich with information when analyzed in the context of fluid, electromagnetic, and chemical models. Obtaining measurements with higher spatial and temporal resolution is clearly advantageous. For instance, measurements obtained with a new electronically-steerable incoherent scatter radar (ISR) present a unique space-time perspective compared to those of a dish-based ISR. However, there are unique ambiguities for this modality which must be carefully considered. The ISR target is stochastic, and the fidelity of fitted parameters (ionospheric densities and temperatures) requires integrated sampling, creating a tradeoff between measurement uncertainty and spatio-temporal resolution. Spatial statistics formalizes the relationship between spatially dispersed observations and the underlying process(es) they represent. A spatial process is regarded as a random field with its distribution structured (e.g., through a correlation function) such that data, sampled over a spatial domain, support inference or prediction of the process. Quantification of uncertainty, an important component of scientific data analysis, is a core value of spatial statistics. This research applies the formalism of spatial statistics to the analysis of Earth's ionosphere using remote sensing diagnostics. In the first part, we consider the problem of volumetric imaging using phased-array ISR based on optimal spatial prediction ("kriging"). In the second part, we develop a technique for reconstructing two-dimensional ion flow fields from line-of-sight projections using Tikhonov regularization. In the third part, we adapt our spatial statistical approach to global ionospheric imaging using total electron content (TEC) measurements derived from navigation satellite signals.
Thesis (Ph.D.)--Boston University