Uncertainty Quantification in Assimilative Mapping of Geospace Observations

Data assimilation is a powerful approach to overcoming the limitations of a given geospace monitoring system. A central concern of assimilative mapping of geospace observations is to account for the role of uncertainty in constructing complete self-consistent maps of high-latitude ionospheric electrodynamical states (e.g., ionospheric electric fields and currents, FAC, Hall and Pedersen conductance and Joule heating) from multiple types of high-latitude geospace observations. This paper illustrates how uncertainty in observations and first-guess models is projected to the assimilative analysis and its measure of uncertainty. The goodness-of-fit of assimilative analysis to independent data can be assessed by cross-validation, which in turn helps calibrate the parameters and assumptions (e.g., a priori ionospheric conductance distributions) adopted in the data assimilation procedure.