Statistical modeling:
The final project of this theme involves analyzing data with some of the
methods discussed in class, and submitting a write-up (about 20-30
pages, double-spaced, not counting figures). The project is due
before the last class.
The data can be from your area of research (though you should discuss
this with the instructor first), or you could use data from the tropical
Pacific as provided. The SOI and the SST for the regions Nino1+2,
Nino3, Nino3.4 and Nino4 are suitable for time series analysis
(e.g. correlation, regression, spectral analysis and filtering). The
equatorial SST and SLP fields are suitable for multivariate time series
analysis (e.g. principal components, canonical correlation analysis,
singular spectrum analysis, principal oscillation patterns, neural
networks). The neural network codes for nonlinear principal component
analysis and nonlinear canonical correlation analysis will be
provided if you are interested in them.
Useful datasets for the project are given below: ( Schematic diagrams
for data description )
SOI (Southern Oscillation Index)
SST for Nino 1+2 (0-10S)(90W-80W), Nino 3 (5N-5S)(150W-90W),
Nino 4 (5N-5S)(160E-150W), Nino 3.4 (5N-5S)(170-120W)
How to read the SST & SLP data below:
Equatorial SST data
Equatorial SLP data
READ Equatorial SST Data
READ Equatorial SLP Data
Optimal modeling - Data Assimilation:
This project is an extension to lab6, further exploreing the applications of
ensemble-based data assimilation methods. It can include, not limitted, such
issues as advanced algorithems of EnKF, for example,Extended KF, square-root EnKF,
inflation factor of EnKF, SPKF etc.. and comparing and analyzing these methods.
You also can use another dynamical model that you may interseted in to perform
EnKF or other assimilation methods. It is expected that a series of experiments
can bring some interesting results.