Reduced dimension models for weather forecasting
This is the subject of my MScR thesis (Master of Science by Research). I did it in Aston University, Birmingham, UK, in 2000.
Here is the abstract:
Weather forecasting is one of the most computationally intensive activities that is routinely undertaken. This project studies the possibility of reducing the dimensions of the models and data sets considered, while maintaining reasonably good predictions.
Techniques dealing with the two problems separately, dimension reduction and forecasting, are applied on real data provided by the European Center for Medium-Range Weather Forecasting, and on an artificial data set generated using the Lorenz equations.
A new algorithm is presented as an extension of the principal interaction patterns framework to neural networks, allowing a simultaneous optimization of the subspace basis for the data projection and the model considered to make predictions.
Advantages and drawbacks of those methods are discussed, and conclusions are drawn from this study regarding the feasibility of reducing the dimensions in the forecasting problem.
And my conclusion, a few years after, is that the method used in not very efficient and doesn't satisfy me anymore. Well, this is research! We do something new... and the most important is to do it correctly, so we can draw conclusions.
You can download the thesis in PDF (1.2 Mb). Please contact me if for any reason you cannot read it, or just for informations, comments, complaints...
The latest source code for the experiments carried on during this project is also available (14Kb, compressed). These programs are written in C++, and make use of CheapMatrix, the library I created for the occasion.
The real data itself is subject to a non-disclosure agreement with the British Atmospheric Data Center, but can be obtained freely for genuine non-commercial research purposes. Alternatively, you can download the artificial data (2.3 Mb, compressed) generated as described in the thesis.