Estimation of the mean function and the principal components of spatially distributed functional data
Piotr Kokoszka, Utah State University
Wednesday, January 19, 2011
3:00 p.m., room 223, Weber Bldg
In many fields, most notably in environmental science and geosciences, but also in medicine and public health, data have the form of temporal curves available at unevenly distributed spatial locations. It is often of interest to combine such data to evaluate global or regional temporal trends. The research I will present is motivated by records of space physics data available at globally distributed observatories and reaching back into the 1960's. These data have been used to evaluate the hypothesis of "global cooling'' in the ionosphere. We cast such data structures into the framework of functional data analysis, and show how the spatial dependence can be used to estimate their main parameters: the mean function and the functional principal components. We conclude with a brief discussion of asymptotic results.