|
Activity Number:
|
185
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Statistics and the Environment
|
| Abstract - #309357 |
|
Title:
|
Statistical Approaches to El Niño Forecasting
|
|
Author(s):
|
Souparno Ghosh*+ and Amanda S. Hering and Salil Mahajan and Marc Genton and Mikyoung Jun and Bani Mallick and Ramalingam Saravanan
|
|
Companies:
|
Texas A&M University and Texas A&M University and Texas A&M University and University of Geneva and Texas A&M University and Texas A&M University and Texas A&M University
|
|
Address:
|
415 college main street, College Station, TX, 77840,
|
|
Keywords:
|
ENSO ; MEOFs ; Vector Autoregression ; Model Selection ; Bayes Factor ; Credible interval
|
|
Abstract:
|
Predicting the evolution of climate on timescales of a season to a few years is challenging. The numerical approach tackles this problem using a supercomputer to solve the physical equations that govern the time evolution of the climate system. The statistical approach involves fitting empirical models to historical data. We construct a family of statistical models focusing on short-term forecasts of El Niño. Three types of models are fitted using Bayesian and a Classical methods: a simple VAR model; a seasonal monthly model; and a seasonal monthly model fit with a window of three months of data. We investigate the performance of these three models trained on monthly data during the period 1960--96 and tested during the period 1997-04.The statistical approach remains competitive with the numerical approach for El Nino forecasting, but is limited by the shortage of historical data.
|
