Analyzing the Diurnal Cycle by Bayesian Interpolation on a Sphere for Mapping GNSS Radio Occultation Data

Abstract Bayesian interpolation has previously been proposed as a strategy to construct maps of radio occultation (RO) data, but that proposition did not consider the diurnal dimension RO data. In this work, basis functions are extended into domain cycle, thus enabling monthly mapping data in synoptic time and analysis atmospheric tides. The spherical harmonics multiplied by sinusoids cycle up arbitrary harmonic degree harmonic. requires regularizer impose smoothness on fits it produces, thereby preventing overfitting formulation for is most probable values parameters determined. Special care required when obvious gaps sampling known occur order prevent false detection statistically significant high-degree atmosphere. Finally, work probes ability generate valid uncertainty fit. postfit residuals dominated measurement noise unresolved variability atmosphere, which nonuniform across globe, violating central assumption interpolation. problem ameliorated constructing using partially resolve temporal approximately every 3 days