|
Hierarchical Statistical Analysis of Global and Regional Environmental
Data
|
|
Observed SST Anomaly 10/98
|
Prediction of 10/98 from 3/98:
Probability-weighted
combination
|
|
|
Based on Berliner, L.M., Wikle, C.K. and Cressie,
N. (2000).
Long-Lead Prediction of Pacific
SSTs via Bayesian Dynamic Modeling.
Journal of Climate, 13,
3953-3968.
|
|
Project Description
|
|
Remotely sensed environmental data are massive in size and are expected
to show smooth variability both in space and time (spatial-temporal
correlation). It is of great importance to use such data effectively
and quickly to infer and predict various aspects of the earth's
ecosystem. One aspect of our research is to fit Bayesian hierarchical
models to regional climate data. We use sea-surface temperature data
in the tropical Pacific to give long-lead forecasts of El Nino/La Nina
events. Another aspect of our research is optimal spatial smoothing of
Total Ozone Mapping Spectrometer (TOMS) data that was collected by the
polar-orbiting Nimbus-7 satellite.
|
|
Objectives
|
|
|
The first objective is to implement hierarchical spatio-temporal
statistical modeling in the processing of climate data and
remote-sensing data. Since the data are in general massive, another
objective is to develop different versions of the statistical models
that explore the usual compromise between computational efficiency and
model complexity.
|
|
Approach
|
|
|
Some progress has already been made on the two objectives outlined
above.
Hierarchical statistical modeling has an immediate use in regional, national, or
even global data-collection programs. A Bayesian approach allows
prediction of one climate variable based on another that is
correlated with it and lagged in space and time.
Regarding the global scale, we intend to develop multi-resolution
spatio-temporal statistical methodology for remote-sensing data from
NASA's Earth Observing System (EOS) initiative. This will be done
in consultation with Dr. Ralph Kahn and Dr. Amy
Braverman, at NASA's Jet Propulsion Laboratory.
|
|
Investigators
|
|
|
|
|
Links
|
|
|
|
