The Geophysical Statistics Project
at The National Center for
Atmospheric Research
Mark Berliner
Department of Statistics, Ohio State University and National Institute for
Statistical Science
Four years ago the National Science Foundation awarded a multi-year
grant to The National Center for Atmospheric Research (NCAR) in Boulder,
Colorado to establish a Geophysical Statistics Project (GSP).
I was fortunate to serve as Project Leader for the past two years.
There are two main topics in this talk. First, I will describe a bit about
what NCAR is, and then talk about GSP and its goals and operation. This
will include a brief review of a variety of research projects we worked on.
Second, I will describe a collection of projects under the general theme
of hierarchical Bayesian space-time modeling using atmospheric and
oceanographic data. Examples are presented for estimating sea-surface winds
employing (i) satellite data (scatterometers); (ii) coarse, global
estimates (ECMWF or NCEP analysis fields); (iii) data of different types
(pressure data); (iv) physical reasoning, leading to forms of the
statistical models (and priors on parameters)--If you do not know what the
things in parentheses are, don't worry. There will not be much presented
on detail; emphasis will be on the notions.