Spatial multi-taper spectrum estimation for nuclear reactor modelling

Multi-taper univariate and cross-spectral analysis is used to investigate the structure of spatial variation in the temperatures measured across the surface of a nuclear reactor. The construction of the spatial tapers over the approximate circular reactor surface is described, along with derivation and sampling properties of the spectral estimates. It is shown how the analysis can be used to identify regular patterns in the temperature measurements, and the spatial transfer function between the reactor temperatures and fuel characteristics. The results are used to specify a parametric regression model for the temperature measurements.