Statistical Methods for Spatio-Temporal Interactions and Their Assessment

Monday, 17 February 2014
Columbus IJ (Hyatt Regency Chicago)
Michael Stein , University of Chicago, Chicago, IL
There are large literatures on statistical methods for processes observed over
time (time series analysis) and processes observed over space (spatial
statistics). Proper statistical methods for environmental processes in
space-time require more than just combining approaches from time series
and spatial statistics, particularly on time scales on which dynamics of the
process are relevant, yielding complex spatio-temporal interactions. This
talk discusses some approaches for addressing these challenges, including
the development of models, estimation methods for the parameters of these
models and diagnostics for assessing the appropriateness of spatio-temporal
models. These issues will be explored in the context of measurements taken
from sun-synchronous polar-orbiting satellites, which are commonly used for
environmental monitoring and which create specific difficulties related to
the fact that observations from such a satellite are concentrated near noon
local time.